SBF did terrible acts from many different moral viewpoints, including that of consequentialism. In addition to those he directly harmed, he harmed the EA movement.
However, from review of what I have read, it seems as if he acted from a sincere desire to better the world and did so to the best of his (quite poor) judgment. Thus, to me, his punishment is a tragedy, though a necessary one. From a matter of ultimate culpability, I don’t know if I would judge him more harshly than the vast majority of people in the developed world: those having the capability to save or dramatically better the lives of people in the developing world, but decline, or those who thoughtlessly contribute to the torture of animals through their participation in the animal product economy.
I wish him comfort and hope that he can find a wiser path forward with the remainder of his life.
However, from review of what I have read, it seems as if he acted from a sincere desire to better the world and did so to the best of his (quite poor) judgment.
Although none of us can peer into SBF’s heart directly, I think a conclusion that he acted from mixed motives is better supported by the evidence. It would take a lot to convince me that someone who was throwing money around like SBF on extravagances (or a $16.4MM house for his parents) was not motivated at least in considerable part by non-benevolent desires.
If one thinks he viewed luxuries bought as part of his fraudulent enterprise as ultimately altruistic because what makes SBF productive --> makes FTX richer --> makes the world better through philanthropy, then we have a framework under which it is impossible to categorize his motives because any behavior can be recast as ultimately motivated by altruism. The base rate of fraudsters being motivated by personal gain is very high, so—unless there’s a clear way to verify lack of personal-gain motivation—I think assuming its absence is doubtful.
I think a conclusion that he acted from mixed motives is better supported by the evidence.
I disagree, but it obviously depends what exactly we’re discussing.
Was his judgment for not coming clean when things were only starting to get bad compromised by not wanting to lose his influence, money, and reputation? Probably!
However, do I think he made some of his most consequential decisions to a significant degree because he thought he could get nice things for himself that way? I actually don’t think so!
Making big decisions for reasons other than impact would ruin his score in the game. It looks to me like his primary drive was optimizing his life for impact like a video game, trying to score the most points. Making a big life decision for self-oriented reasons would be forfeiting points he could otherwise have gotten, which would feel deeply unsatisfying for someone who’s always thinking about how to get the highest score.
Also, $16 million is peanuts for someone who’s personally worth $2-10 billion, which SBF probably “”somewhat reasonably”″ thought he was worth when things were going well (or when he optimistically thought he could make them well again soon enough). It’s like what a couple hundred bucks is to someone with a net worth of $80,000 – not that significant! Imagine you’re making decisions for tens of millions every day. I can’t emphasize enough how unimportant it then becomes whether you pay 1 million or 16 million for your parents’ house (which you may or may not also plan to use for yourself or some of your coworkers/flatmates). Just like with offices or the building where he planned to sleep in, the price isn’t the relevant variable (what’s more relevant is, “Does it take up your time or people’s time with high opportunity costs? Is it close to the office where you work? Is it convenient?” Etc.)
I feel like,
if someone displays as much single-minded obsessiveness as SBF,
and his conversations with the people around him tell you he’s hyperfocused on impact (he wasn’t just obsessing about how to grow FTX for money, he was also constantly discussing ideas around grantmaking strategies or paying Trump to not run again for election),
and his close social circle is reinforcing the impact-orientedness,
and there’s a credible ideology behind it which generated many other examples of similarly impact-obsessed people (hopefully the vast majority of them without the recklessness and “no fear of bad consequences” part),
then I’d say the evidence points to being primarily impact-driven pretty much as strongly as it gets. Of course, no one has the motivations of a perfect utilitarian robot, but I think SBF was trying, so it doesn’t seem fair to call it “acted from mixed motives.”
Edit to add: We could discuss whether he was “unusually biased” for someone who’s trying to have a ton of impact – hopefully yes (otherwise EAs don’t stand a chance of getting impact right), but that’s still meaningfully different from “acting from mixed motives.” (Also, I’d say the most severe bias was probably a blindspot around risk taking and risks of things going badly, rather than e.g., being too much into a rich lifestyle or being too much into fame or whatever. FWIW, my best guess is that he also was too much into wanting power/influence; I’m just saying that it doesn’t seem like the primary reason things went so poorly.)
Edit2: Somewhat relevant comment I made a year ago about why I think SBF having dark triad traits is compatible with the above picture.
Also, 16 million is peanuts for someone who’s personally worth 2-10 billion
If you’re going to compare outflows to guess at motivation, I think it’s better to use actual charitable giving numbers as the comparator rather than perceived wealth. Doubtless underinclusive on both numbers, but IIRC the grants through FTXFF were ~$150MM and the outflows to his parents alone were ~$25MM (there was also ~$10MM in cash). That doesn’t suggest the self-serving cashflows were peanuts in either a relative sense or an absolute sense.
If you assumed based on past actions that ~ 90% of the money was going to end up donated to charity, then ~$1B of the loss found by the district court today was attributable to the 10% that wasn’t. Given SBF’s crimes, and the fairly brazen nature of his perjury at trial, I would credit his observed actions over what he said he’d do with all the money in the end.
That’s a good point. If there weren’t a convincing story for why more donations weren’t made or at least set up to be made soon, I’d say your point counts for quite a lot!
However, in this specific case, I feel like there are good reasons why I wouldn’t expect that many donations to be made right away:
FTX did have a hole in the bank! It’s interesting that donations were made at all given that they were strictly speaking insolvent. (Of course, he had to keep up appearances or pay for previously-made commitments, etc., so it’s not too surprising. I’m just saying it probably wouldn’t have been wise even for someone as risk tolerant and unconcerned-by-the-illegality-of-it as SBF to donate much more when there was still a hole in their wallet.) (Edit: Admittedly, this point also cuts against giving money to his parents.)
Longtermist grantmaking (which SBF thought was most impactful, I believe) already started to feel a bit crowded once FTX Future Fund came in, so it’s not actually that easy (or sensible) to deploy $100s of millions on short notice in that cause area.
The FTX Future Fund was still relatively new; it makes sense to scale up giving as you learn things and do various types of preparatory work for your grantmaking/”active grantmaking.”
“Donating now vs later” isn’t actually the most obvious of EA strategy questions, esp. if you think you’re greatly beating the market on investment returns, which he had been doing at least in his own mind.
SBF’s entire philosophy was about massive wins and game-changing ambitious stuff, so he probably thought of the donations he had already made or was making at the time as not that significant compared to what he was gonna do later, and probably had tentative ideas for long-term plans like “amass enough money to buy a chip company and then use that as leverage to get more AI safety work done at labs,” or something super ambitious like that, for which it felt worthwhile to keep investing in the growth of his FTX empire.
Lastly, and somewhat related to the previous bullet point of weird super-ambitious ideas, he did discuss the paying Trump thing and, at least according to Michael Lewis’s sources, he was entertaining the thought (though probably not super seriously because this was when they had a massive hole in the bank?) of paying a sum for it that would have been a lot bigger than the $150 million in charitable contributions you mentioned. (But sure, you can say that, since nothing happened there, it was only talk. We don’t know, but I find it plausible that he’d have liked the thought of being the guy who prevented Trump from running!)
Edit: If someone says “giving 10 million to his parents (and a house, but we’re not sure if the house had other uses) shows he must have had mixed motives,” I’d be like: Okay, yeah, it does look like this isn’t what a utilitarian robot would’ve gone for, but when I hear “mixed motives,” I think of something like 50-50 or at least 40-self-oriented, 60-altruism, whereas giving money to get your parents to retire early could also be compatible with something like 10-90 (or even 5-95), “focus on impact for the big decisions, but do something nice for your parents once you’ve made it big.”
SBF likely had mixed motives, in that there was likely at least some degree to which he acted in order to further his own well-being or with partiality toward the well-being of certain entities (such as his parents). The reasoning that you mentioned above (privileging your own interests instrumentally rather than terminally such that you as an agent can perform better) is a fraught manner of thinking with extremely high risk for motivated reasoning. However, I think that it is one that serious altruists need to engage with in good faith. To not do so would imply giving until one’s welfare was at the global poverty line, which would probably impair one too much as an agent. Of course, I’m not saying he was engaged in good faith regarding this instrumental privileging argument, but I cannot preclude the possibility.
Regardless, I have been persuaded by everything that I have seen that a significant part of SBF’s motivations were to help advance a world of higher well-being. Of course, from a deontological perspective he did wrong by his dishonest and fraudulent actions. From a consequentialist perspective, the downside risks had such incalculable costs that it was terrible as well.
But the sincere desire of his to make the world a better place makes me sympathetic of him in a way that I probably would not be with similarly sentenced other convicts. Given a deterministic or random world, I understand that all convicts are victims too. But I cannot help but feel more for one who was led to their crime by a sincere desire to better the world, than say, to kill their spouse in a fit of rage, or advance themselves financially without any such altruistic motivation.
This is just not true if you read about the case, he obviously knew he was improperly taking user funds and tells all sorts of incoherent lies to explain it, and it’s really disappointing to see so many EAs continue to believe he was well-intentioned. You can quibble about the length of sentencing, but he broke the law, and he was correctly punished for it.
Please note that my previous post took the following positions:
1. That SBF did terrible acts that harmed people.
2. That it was necessary that he be punished. To the extent that it wasn’t implied by the previous comment, I clarify that what he did was illegal (EDIT: which would involve a finding of culpable mental states that would imply that his wrongdoing was no innocent or negligent mistake).
3. The post doesn’t even take a position as to whether the 25 years is an appropriate sentence.
All of the preceding is consistent with the proposition that he also acted with the intention of doing what he could to better the world. Like others have commented, his punishment is necessary for general deterrence purposes. However, his genuine altruistic motivations make the fact that he must be punished tragic.
All punishment is tragic, I guess, in that it would be a better world if we didn’t have to punish anyone. I guess I just don’t think the fact that SBF on some level “believed” in EA (whatever that means, and if that is even true) - despite not acting in accordance with the principles of EA—is a reason that his punishment is more tragic than anyone else’s
As others have observed, it’s been taboo to publicly speculate on SBF’s innocence since long before his conviction, including—or perhaps especially—in EA spaces. So please excuse the pseudonym, but...
I think this sentence is a tragedy. This isn’t the place to get into my general issues with the United States incarceration system, but even putting that to one side, I believe there’s much more to this story than meets the eye and it would be nice to see more EAs seriously grappling with the arguments in his defense.
I think people who have followed this unusually closely should be encouraged to argue for what they think is right if they have a strong take, but I just don’t think this theory is likely. An innocent person would be more likely to talk more freely about things/be less evasive and they’d probably have a better explanation of how it is that they could have missed an 8 billion hole in the bank. It’s suspicious if you need to make the same move (“he could’ve not seen this” or “he could have not looked closely at that”) multiple times to preserve the chance of innocence. Every time you use a not-too-likely excuse like this, your hypothesis takes a hit.
I don’t have a strong view on the sentence length. I think the main reason for a low sentence length (compared to the guidelines for the counts he was convicted of) is that I do believe SBF was strongly altruistically motivated, which is unusual for cases like that. I think the main reason for a long(-ish) sentence length is perjury and evasiveness.
Pleading not guilty is not perjury. For starters, it isn’t a claim of innocence, merely a demand that the prosecutors put up or shut up. Second, perjury has a specific definition, which includes that a material, false statement was made under a legally valid oath or its equivalent.
SBF testified under oath in his own defense at trial. He was not obliged to testify, and the jury could not draw a negative inference from his decision not to. In fact, most US defendants do not testify.
If he made a material, false statement with knowledge of its falsity under oath at trial, then he committed perjury. Testifying under oath that you “don’t remember X” when you know you actually do is perjury (assuming that X is material, a rather low bar to meet).
Example (quoting the Government’s sentencing memo):
Bankman-Fried’s minimization and false explanations concerning such conduct was pervasive during his testimony. As one example, he claimed that he first learned that Alameda had a roughly $8 billion fiat liability to FTX in October 2022. (Tr. 2523-24). That was a lie, as is evidenced by the testimony of Yedidia, Ellison, Wang, and Singh, all of whom testified that they discussed Alameda’s fiat liability to FTX with Bankman-Fried months earlier. (Tr. 167, 173-74, 436-37, 439-40, 619-20, 769-71, 1346, 1348, 1359; GX-50). The documentary evidence also proved that Bankman-Fried was not telling the truth during his testimony. Government Exhibit 50, for example, a June 2022 spreadsheet which was viewed by the defendant, showed a roughly $11 billion fiat liability owed by Alameda to FTX. (GX-50).
I agree that some of his behavior was just unproblematic “being very literal about answers.”
But the thing I mean by evasiveness was more stuff like:
Not remembering important things
Not giving answers that substantially clarified what happened
Saying “I don’t know which piece of evidence you’re referring to?” when he was several times asked simple questions on whether he did or didn’t do something. Note that this is the opposite of taking questions too literally; instead, it’s being deliberately obtuse to mask his refusal to ever disclose information that the prosecution didn’t already present, which is a central example of “being evasive.”
Regarding perjury, a lawyer (I believe it was an FTX lawyer?) testified that SBF asked him what excuses he could use to explain that the money is gone, then the lawyer suggested some pontential avenues but was like “they won’t work because it’s not actually legal/doesn’t hold up with the details” (paraphrased), and SBF nodded at that, but then used these excuses anyway talking to the media?! And then when asked about what he had said in the media, he didn’t take it back, which at that point constitutes perjury.
I agree that some of his behavior was just unproblematic “being very literal about answers.”
You have my respect for acknowledging that; it’s very rare that anyone does.
Not remembering important things
I really think people should cut him a lot more slack for this. On November 11th, new FTX management said in a public announcement that, “Sam Bankman-Fried has resigned his role as Chief Executive Officer and will remain to assist in an orderly transition”, while behind the scenes they were reporting him to authorities[1]. So he probably didn’t expect that he was about to be permanently shut off from access to internal company data to the greatest extent that new management could manage and therefore he may not have been secretly exporting all the relevant data to his personal storage as may have been prudent. (In fact, he’d been busy turning off autodelete “on any place I found it”.) It’s very hard to remember exactly how things played out when in the middle of a crisis you’re suddenly shut off from almost all the relevant data and it’s now a year or so later.
He also gave dozens of interviews. A very literal person, when asked, “Did you say xyz on this date?” is not going to think, “I expect they’re quoting me...but sometimes they’re just paraphrasing me in a misleading way...and obviously I don’t actually remember...but it’s going to look really bad if I say I don’t remember and then they put a tweet from me on the screen saying it...so I better say yes.”
He was also living off bread and peanut butter in one of the worst jails in the country with limited access to the internet, discovery material and his ADHD meds, he’d just gone through a year of the globe turning on him with the final nail in the coffin being his former girlfriend testifying extensively against him right in front of him, and, I expect, he was doing what he could on the stand to mask autistic mannerisms so that he wouldn’t come across so unlikeable and shady. This can’t have helped his ability to recall things.
Not giving answers that substantially clarified what happened Saying “I don’t know which piece of evidence you’re referring to?” when he was several times asked simple questions on whether he did or didn’t do something.
Note that this is the opposite of taking questions too literally; instead, it’s being deliberately obtuse to mask his refusal to ever disclose information that the prosecution didn’t already present, which is a central example of “being evasive.”
I’d be interested to hear examples (genuinely). I can’t think of times when it seemed like this is what was happening, rather than him just “being autistic” (or not actually remembering, see above). In that link in my last comment it says, for example, “In their article Autism in the Criminal Justice System, Judge Kimberly Taylor and other experts urge criminal justice professionals to ‘[u]understand the need to repeat and rephrase questions” and note that “difficulty in maintaining eye contact or insistence on changing the subject of conversation to a topic of their choice—all typical diagnostic behaviors of a person with autism—can mislead an investigator, attorney, or judge.” Judy Mark—a UCLA Disability Studies faculty member who has an autistic son...acknowledges “answers that seem non-responsive’ and ‘times when [autistic individuals] might repeat themselves or become obsessed on an issue, something we call perseveration. They might become particularly focused on minute details that others find irrelevant, something we call hyperfocus.’”
Regarding perjury, a lawyer (I believe it was an FTX lawyer?) testified that SBF asked him what excuses he could use to explain that the money is gone, then the lawyer suggested some pontential avenues but was like “they won’t work because it’s not actually legal/doesn’t hold up with the details” (paraphrased), and SBF nodded at that, but then used these excuses anyway talking to the media?! And then when asked about what he had said in the media, he didn’t take it back, which at that point constitutes perjury.
Yes, an FTX lawyer[2]. Again, I’d be interested to hear more details. It sounds to me like they didn’t realize the nature of the mistake they’d made until the weeks leading up to the run on the exchange and I expect they thought that, given their ignorance at the time, they had not committed fraud. But I also expect that now they were facing a liquidity crisis—and didn’t think anyone would believe them that it was an accident—they were keen to know if it would have been fraud even if they had known exactly what they were doing. I can’t think of a time he lied in the media on the basis of this (or, ever).
In fact, Nishad appears to slip up at one point on this: “However, not all elements of the prosecution narrative line up neatly. Singh said he left the crucial June meeting still thinking things were OK and did not realise customer funds were being raided until September” (Financial Times).
I’m getting almost all of this from the first website I linked to, by the way.
“On November 9, 2022 . . . S&C attorneys in our Criminal Defense & Investigations Group, in consultation with Mr. Miller, reported the concern to federal authorities, including the United States Attorney’s Office for the Southern District of New York [USAO], the Securities and Exchange Commission and the Commodity Futures Trading Commission.” (from this paper)
I don’t want to spend too much time on this so won’t answer to all points, but I wanted to point you to some examples for this bit about evasiveness by saying things like, “I don’t know what this is referring to”:
I’d be interested to hear examples (genuinely)
See the transcript here: the word “referring” occurs 30 times and at least a couple of those times strike me as the weasel-like suspicious behavior of someone whose approach to answering questions is “never admit to anything unless you learn that they already have the evidence.” So, he always answers first with “not sure/don’t know what you refer to/don’t remember” and only admits to things when shown evidence.
This behavior is strikingly abnormal and different from how a person who doesn’t have anything to hide would behave.
(Edit – and again, it seems to me like it’s different from autistic literal-mindedness! Literally answering the question would mean to comb your memory and answer without regard for what the prosecution is referring to. It would also include saying confidently “no” if you’re sure you never said something.)
Someone trustworthy would answer questions immediately, sometimes admitting to things that the prosecution may not already know.
Some examples:
Q. You also marketed FTX as a safe crypto exchange compared to your competitors, didn’t you?
A. With FTX US I think that may be the case. I am not sure about FTX International.
Q. Did you or did you not market FTX International as safe compared to other crypto exchanges?
A. I don’t specifically remember that. I am not sure.
MS. SASSOON: If we could pull up Government Exhibit 900
A. The government offers Government Exhibit 900
THE COURT: Hearing no objection, it’s received. (Government Exhibit 900A received in evidence)
MR. COHEN: I thought it was in already.
THE COURT: No harm, no foul.
MS. SASSOON: I believe the full video is in. This is just a screenshot. Mr. Bianco, if you could publish that, please. We can go ahead and take that down.
Q. You publicly described FTX as the most regulated crypto MBAN3 exchange by far, didn’t you?
A. I think that’s right.
Q. And you also acted like you cared about customer protections, right?
A. I think I did care about them, yes.
Q. And you made public statements to that effect, didn’t you?
A. I probably did.
Q. I didn’t hear you.
A. I probably did.
Q. Yes or no, do you recall making statements that you cared about customer protections?
A. Yes.
Q. In fact, over and over again in public forums you described FTX platform as safe, correct?
A. I am not sure specifically what that is referring to. I may have.
Q. Yes or no, do you recall making numerous public statements to the effect that the FTX platform was safe?
A. I recall with respect to FTX US. It may be true with respect to FTX International, but I don’t specifically recall, no.
Q. You were CEO of FTX International, yes?
A. Yes.
Q. Sitting here today, you cannot recall one way or the other whether you made public statements that FTX was a safe MBAN3 platform? 1
A. I am not sure exactly what you are referring to. I made a lot of public statements.
Q. Yes or no, do you recall making public statements that FTX was a safe platform?
A. I can’t think of a specific one off the top of my head.
Q. Generally, do you recall in substance making statements that FTX was a safe platform?
MR. COHEN: Objection.
THE COURT: Overruled.
A. Some things that were sort of like that, yes. I am not sure exactly what you are referring to. But I am not saying —
THE COURT: Mr. Bankman-Fried, the issue is not what she is referring to. Please answer the question.
Q. Putting aside what I’m referring to, I’m asking whether you recall making statements as CEO of FTX that in substance stated that the FTX platform was safe.
A. I remember things around specific parts of the FTX platform that were related to that. I don’t remember a general statement to that effect. I am not sure there wasn’t one.
Q. In media interviews isn’t it true that you insisted on that FTX had protections for retail customers?
A. Yup.
Q. You told your customers that users’ funds and safety come MBAN3 first, didn’t you?
A. Something to that effect, yes. 2
Q. And you also made statements that you would always allow withdrawals, didn’t you?
A. Yup.
MS. SASSOON: If we could pull up what’s marked as Government Exhibit 829. The government offers Government Exhibit 829.
MR. COHEN: No objection.
THE COURT: Received. (Government Exhibit 829 received in evidence)
MS. SASSOON: Mr. Bianco, can you publish that.
Q. Mr. Bankman-Fried, can you read the first line of your tweet from August 9, 2021.
A. Sure. And, as always, our users’ funds and safety come first.
Q. Beneath that do you see where it says, we will always allow withdrawals except in cases of suspected money laundering/theft/etc.?
A. Yup.
MS. SASSOON: We can take that down.
[...]
Q. You also claimed that FTX had a conservative approach to managing risk, didn’t you?
A. OnóóI’m not sure exactly what that was referring to.
Q. You don’t recall saying that?
A. I don’t remember the context.
Q. Do you recall saying that in any context?
A. I’m not confident.
[...]
Q. So is it your testimony that as CEO of FTX, after this catastrophic event, you did not learn the details of the code change that you directed?
A. That’s correct. I trusted Gary and Nishad.
Q. You testified on direct that FTX had an AWS database, correct?
A. Yup.
Q. And you described its content, right? For example, it stored customer account information?
A. Yup, that’s right.
Q. And it had information about trades?
A. Yup.
Q. And customer balances?
A. Yup.
Q. And as CEO, you had access to the database, correct?
A. Nope.
Q. Your testimony is that you did not have the ability to access the database?
A. I never did. To my knowledge, I didn’t have access to it.
Q. I’m asking you whether you had authorization to search the database.
A. I have no idea whether someone had created an account in my name that in theory was designed for me. If so, I never used it.
Q. And so it’s your testimony that until October 2022, you never looked in the database.
A. That’s correct. And even as of then, I never looked in the AWS database.
Q. After FTX declared bankruptcy, isn’t it true that one of the first things you did was try to restore your administrative access to the AWS database?
A. That’s not how I would put it.
Q. Isn’t it true that in the weeks following the bankruptcy, you asked to have your access to the AWS database restored?
A. NotóóI was not specifically looking for my personal access to the AWS database.
Q. Isn’t it true you were requesting AWS access?
A. I was requesting it on behalf of the joint provisional liquidators in the Bahamas.
Q. So yes or no: You made requests to restore access to the AWS database?
A. I’m not sure exactly what you’re referring to here.
THE COURT: Look, could you just answer the question instead of trying to ask the questioner what she’s referring to? THE WITNESS: Okay.
A. No.
Q. Isn’t it true that you made to-do lists after FTX’s 4 collapse that included things like “try to get AWS access”?
A. Probably.
Q. And so isn’t it true that you were trying to get AWS access after FTX declared bankruptcy?
A. Yes.
To me, the focus on “What this is referring to” is illuminating because it shows how SBF is laser-focused on what the prosecution has on him. What’s strikingly absent is a desire to try hard at remembering so he can tell as much of the truth as possible.
Thanks for going to the trouble of sourcing these quotes.
To me, the focus on “What this is referring to” is illuminating because it shows how SBF is laser-focused on what the prosecution has on him. What’s strikingly absent is a desire to try hard at remembering so he can tell as much of the truth as possible.
I mean, in the “practice” cross-examination, I think he was frequently rebuked for guessing at what the prosecutor meant or sharing information beyond a short, binary, direct answer to the question the prosecutor (thought she) was asking. For example, “A. And by “go negative,” you’re talking about negative in a particular coin or negative net asset value? Q. Just have a negative balance, Mr. Bankman-Fried. A. Sorry. I— … THE COURT: … I’ve gotten beyond my tether here. … part of the problem is that the witness has what I’ll simply call an interesting way of responding to questions for the moment. Q. Mr. Bankman-Fried, in May of 2022, were you aware that account ID 9 @AlamedaResearch.com could have an overall negative value? A. I am giving you my best guess at answering the question. Q. I’m not asking for a guess. I’m asking what you understood at the time. A. I am going to answer what I think the question you are asking is, but I apologize if I’m answering the wrong question.” I think the precise question she intends to ask here is extremely important, but she won’t specify (or doesn’t understand that she needs to.) It’s a very complicated case and the details matter. If Alameda’s trading accounts collectively had a net liability of billions to FTX for an extended period with no uncorrelated collateral, then they “borrowed” billions here in the way ~everyone thinks they did. But if one subaccount was temporarily negative $3 billion in a particular coin while another subaccount was temporarily positive $4 billion in another coin and it’s mostly for market-making purposes, then the sense in which Alameda was “borrowing” $3 billion via its customer accounts is extremely different.
And then there are plenty of reports of how when SBF gave extra information that wasn’t asked for, that was surely “evading questions, trying to pour forth verbiage to distract Sassoon from what she’d asked.” (from here)
Perhaps we’re understanding the term “referring” differently—I hear it as “I don’t understand exactly what you’re asking me (and if I guess wrong you might make me look like a liar)” while you might be hearing it as “I don’t know which of the many pieces of supporting evidence have prompted you to ask this question and I don’t want to reveal more than you already know about.”
Taking the first example of “referring to” in your comment, we have:
Q. In fact, over and over again in public forums you described FTX platform as safe, correct?
A. I am not sure specifically what that is referring to. I may have.
I’m picturing his thought process as something like, “I think I used the word ‘safe’ multiple times in public forums in reference to FTX U.S. If that is what she’s referring to then my answer is ‘Probably.’ But maybe she’s referring to that tweet someone else wrote using the FTX Twitter account that said FTX International was safer than Binance. And I can’t think of any times I’ve actually described FTX International with the word ‘safe’ and I don’t think I would have said that, so what do I say? If I say ‘Yes’ then I’m giving the impression that I said something bad that I don’t think I said, but if I say ‘No’ then she might bring up that tweet and make me look like a liar.”
I think that’s how my mind would respond anyway, given the way other interactions had gone. For example, “Q. Now do you remember telling Zeke Faux in early 2022 that Alameda played by the same rules as other traders? A. Not in that wording, no. Q. So you don’t recall that. A. No. Q. Do you recall telling him that in other wording? A. I recall saying that Alameda wasn’t front running other customers, that its trading access was like other customers.” If SBF had given a straight “no”—because he did not in fact make the claim that the prosecutor is heavily implying he did—and then the prosecutor had shown the jury a quote from Faux paraphrasing SBF that was taken out of context, would that have been perjury? I would definitely have been wary of giving yes/no answers to imprecise or misleading questions.
It would also include saying confidently “no” if you’re sure you never said something.
Who is ever sure they never said something? I’m not sure why you and I disagree here. I feel like the “normal” way to answer questions about things you may or may not have said a year or so ago is to say “yes” when you’re >50% sure you said it and “no” otherwise. Or, perhaps when testifying in your own defense, to just say “yes” every time you get the sense the prosecutor is about to pull out some written record of you saying it, even when that’s technically lying because you don’t actually remember, because it’s not worth the risk. But I think literal-minded people will feel very uncomfortable having only those two options and would rather say “I don’t recall”, especially if in squeezing their 25% confidence into a “no” only to get proven otherwise would constitute perjury.
I don’t know which of the lines above are the 2+ suspicious ones, but none of them strike me as such.
He was also living off bread and peanut butter in one of the worst jails in the country with limited access to the internet, discovery material and his ADHD meds [ . . .]
I’d note that Judge Kaplan invited the defense to seek a delay in the trial date if it felt that SBF being remanded to jail prejudiced their ability to put on a defense. They did not do so (likely in an attempt to create a longshot appellate issue, in my opinion). He also had access to a lot of discovery material for many months before being remanded.
There’s also a bit of tension between SBF’s claim that he didn’t have access to information needed to prepare his defense, and that the government dumped too much information on his defense team (at least 7.7 million pages, if I read correctly). Now, I don’t know what was in the Brady material, but I’d want specifics before concluding that important information was unavailable to the defense.
That’s interesting. Do you know when that invitation was (I can’t find it)? If it was close to the trial date, fair enough. If it was at the point of being remanded, it looks like he ended up experiencing a lot of problems that weren’t particularly foreseeable.
I can fact check if you’re curious, but just going by memory for now:
In SBF’s “practice” testimony that the judge requested, someone made a comment about how very convenient it was that he wasn’t able to produce the document retention policy that he kept referring to in an effort to defend himself. He replied that he’d requested it from FTX many times.
There was at least one complaint in the run up to the trial that a huge amount of discovery material had been made available to the defense at such a late date, and SBF was having so many problems with Internet access, that this should hardly count as an opportunity to review the information.
There’s also a huge difference between “information the prosecution wants to use against the defendant” and “information the defendant wants to use to defend himself”. I think the basic situation was that SBF didn’t have a realistic opportunity to look at the former before the trial and that he never had access to the latter (except for information that was inextricably part of evidence that prosecutors had judged served their own ends more than SBF’s overall.) I don’t think there’s a tension between millions of pages being used against you and having minimal access to data that might exonerate you.
And when the defendant began complaining that his recent incarceration made trial preparation by that date more difficult for him, the Court offered to consider any request by the defendant for a substantial adjournment of the trial to give him more time in which to prepare. Tr., Aug. 30, 2023 [Dkt 249], at 20:9-21 :25. Yet defendant conspicuously failed to request any additional time.
(Docket order of Sept 12, 2024, citing transcript of August 30). He was remanded on August 11, so it was pretty soon after that. Moreover, any competent defense attorney would know that they could seek a trial delay on this basis; they did not have to wait for Judge Kaplan to broach the possibility himself.
The district court addressed the late production argument in the same order:
Moreover, the recently produced documents are text searchable indexed productions, the majority of which came from the defendant’s own Google accounts and therefore were accessible to the defendant long before the government received those ve1y same materials in a late production from Google. [ . . . .] Third, the premise of defendant’s position here is that he personally is entitled to review and consider every single piece of the discovery in this case, to generate unspecified work product in respect thereof, and so on. But the premise is incorrect. Defendant is represented by a substantial team of extremely able retained lawyers. He proposes to call at trial seven expert witnesses. And no represented defendant, much less a defendant detained pretrial in order to protect the public, has a constitutional or other right to inspect eve1y bit of discove1y. [. . . . .] Finally, as the Court suggested in remarks on August 30, 2023 (Tr., Aug. 30, 2023 [Dkt 249], at 20:20-21: 1), and as remains true today, the defendant has not made any detailed showing as to specific materials that he claims he has been unable to access personally and the reasons why any such personal inability actually would impede his defense. Generalities about millions of pages without connecting those dots simply is not persuasive. Defendant’s team of lawyers is able to provide defendant with any materials that he would like to review on hard drives at the MDC.
(typos in docket text).
“[I]nformation the prosecution wants to use against the defendant” isn’t an accurate characterization of the government’s Brady obligations—see the Wikipedia article linked earlier. “[P]art of evidence that prosecutors had judged served their own ends more than SBF’s overall” is pretty much the polar opposite of what Brady requires. And I think it is rather unlikely that USAO/SDNY would play games with Brady and thereby risk what they saw as a slam-dunk conviction, especially where the defendant was represented by well-bankrolled counsel and had at least some inside knowledge of what documents might be out there.
Also, none of this seems tied to the specific perjury findings the Government asked the district court to consider—that “Bankman-Fried falsely testified that he did not know about Alameda spending customer deposits, that he did not know until late 2022 that FTX’s code features were being improperly used to take customer money, and that he thought in November 2022 that his tweets were accurate and not misleading.” (p. 31-32 here).
Finally, as to evasiveness / hair splitting / dodging more generally, I’d note that Judge Kaplan has been on the bench for thirty years and has plenty of exposure to defendants who have been remanded, witnesses and defendants who may have conditions that affect their ability to give testimony, and those in extremely complex cases with voluminous discovery. I see no reason to believe he was incapable of factoring those things in to the extent relevant, or unwilling to do so, and trust his findings of fact more than SBF’s self-serving narrative unless given a good reason to depart from that.
Thank you—I misread you, I apologize, I thought you’d said the judge offered a trial delay. (Although it’s still somewhat interesting that Judge Kaplan reminded the defense that they could submit a request for a delay and he would consider it.)
Clearly the Court did not find this persuasive, but I still want to share some sense of the difficulties SBF was facing: at “the latest opportunity the defense had to observe whether the plan for access to the internet-enabled laptop would work as promised,” due to a number of factors, SBF “spent an entire day getting access to a single document. That is one day he could not spend working at the MDC or conferring with his lawyers.” (from here). Perhaps Bradyonly requires that access is given to SBF’s lawyers (who were not all “well-bankrolled” by the way—Mills at least worked pro bono), but I still think these kind of things are worth mentioning in discussions of the extent to which SBF’s testimony was evasiveness vs. genuinely not recalling.
Also, even if “the recently produced documents” were mostly “from the defendant’s own Google accounts,” he had at least 2.5 million pages in his Google accounts, so whether they were technically “accessible” is not really the point. (Assuming he did have access—SBF reported that FTX cut off his access to his personal LinkedIn account, for example.)
But I do think I veered into talking about how FTX would only provide data to the prosecution not the defense, when my original comment that started this thread was just the words “discovery material”, which may have confused things.
I see no reason to believe he was incapable of factoring those things in to the extent relevant, or unwilling to do so
I think Judge Kaplan’s apparent dislike of SBF has been widely reported and I assume we’ll have to agree to disagree on whether that biased any of his judgments in this case.
I think Judge Kaplan’s apparent dislike of SBF has been widely reported and I assume we’ll have to agree to disagree on whether that biased any of his judgments in this case.
Yeah, we’ll have to agree to disagree on that.
I don’t have videotape of Judge Kaplan, of course, but some behavior that may come across as “apparent dislike” could actually reflect sound courtroom management techniques. For instance, sternly lecturing an defendant who does things like use a VPN while on tight electronic lockdown is a restrained use of the district court’s power, authority, and responsibility to manage the proceedings before it. It can be an attempt to prod an errant defendant back in line without bringing out the heavier-handed powers (although Judge Kaplan did end up remanding SBF after the second tampering episode, on which he was affirmed by the Second Circuit).
To be perjury, false, material testimony must be under a legally valid oath or its equivalent. Lying to the media isn’t perjury; generally it is not a crime at all.
What I meant with by “he didn’t take it back” is a situation as follows:
The prosecution asks him if he made certain claims in the media. SBF says “yes” or “it appears that way” or whatever. The prosecution at some other point in the trial (maybe days earlier, maybe afterwards) asks some specific details about how FTX accounts were structured and how money was moved that contradicts what SBF said in the media. At some third point in the trial, they ask him if he deliberately lied to the media/gave false accounts about how things worked, and he said no. (Or, instead of asking him “did you tell the truth to the media?,” maybe they just asked him the same question the media asked him, and SBF stuck to his guns to avoid admitting that he mislead the media – similar outcome because he’s now saying the false thing in the trial setting.)
(I’m not actually sure this exact thing happened – it’s been a while since the trial. But I’d guess there were cases like that where you can look at different responses he gave to various questions and find that they’re in tension with things established by the prosecution from other witnesses. That’s probably also what the judge meant when he said that he found three instances of perjury? But I’m flagging that I haven’t looked at Kaplan’s elaborations on those three points. It’s more that I trust that Kaplan is likely to come to the right interpretations on those counts because when I followed the trial, it also felt to me like SBF was, at various points, getting trapped/caught in a web of questions, previous answers, and others’ testimony. In other words, I’m pretty sure I thought at various points “this is probably perjury established, if we believe this other witness?” – but I didn’t write those down, so I don’t remember the exact web of testimonies that would’ve conclusively trapped him.)
He got off easy, in my opinion. I wrote earlier here about the need for general deterrence, as well as enhancement for certain aggravating factors like witness tampering and perjury.
For non-U.S. readers, the sentence may seem pretty harsh; the actual time to serve may end up roughly equal to the median UK single murderer who did not bring a weapon to the scene and whose conduct did not involve any other statutory aggravating factor. But, compared to what the US criminal justice system regularly hands out for various levels of moral culpability and harm caused, he got (and many white-collar criminals get) significantly less than his culpability level and harm caused would predict. I think “rich people crime” getting a significantly less stern response than ordinary crime is damaging to the rule of law and the social contract in this country.
I also feel little pity for him, at least relative to the median US convicted individual. A large number of people who receive significant prison sentences experienced great childhood trauma, suffer from significant mental illness that contributed to their offenses, and had relatively few good options in life. SBF had great privilege and could choose among an extremely broad range of lawful and attractive paths in life. I don’t think anyone has suggested that any mental illness or autism was contributory to his offense.
So in this case, my desire for consistency and inter-defendant fairness trumps the broader concerns I have about the US system being too punitive.
While I see what you’re saying here, I prefer evil to be done inconsistently rather than consistently, and every time someone merely gets what they deserve instead of what some unhinged penal system (whether in the US or elsewhere) thinks they deserve seems like a good thing to me.
(I don’t personally have an opinion on what SBF actually deserves.)
I can’t speak for the disagrees (of which I was not one), but I was envisioning something like this:
You are one of ten trial judges in country X, which gives a lot of deference to trial judges on sentencing. Your nine colleagues apply a level of punitiveness that you think is excessive; they would hand out 10 years’ imprisonment for a crime that you—if not considering the broader community practices—would find warrants five years. Although citizens of county X have a range of opinions, the idea of sentencing for 10 years seems not inconsistent with the median voter’s views. The other judges are set in their ways and have life tenure, so you are unable to affect the sentences they hand down in any way. Do you:
(a) sentence to five years, because you think it is the appropriate sentence based on your own judgment;
(b) impose a ten-year sentence that you find excessive, because it prevents the injustice of unequal treatment based on the arbitrary spin of the assignment wheel; or
(c) split the difference, imposing somewhere between five and ten years, accepting both that you will find the sentence too high and that there is an unwarranted disparity, but limiting the extent to which either goal is compromised.
I would be somewhere in camp (c), while Ben sounds like he may be closer to camp (a). I imagine many people in camp (b) would disagree-vote Ben, while people in camp (c) might agree, disagree, or not vote.
I can think of grounds to disagree, though. Say for example you were able to disproportionately protect e.g. white people from being prosecuted for jaywalking. I think jaywalking shouldn’t be illegal, so in a sense any person you protect from prosecution is a win. But there would be indirect effects to a racially unfair punishment, e.g. deepening resentment and disillusionment, enabling and encouraging racists in other aspects of their beliefs and actions. So even though there would be less direct harm, there might be more indirect harm.
I think the indirect harms are at work in this case too, and it’s just a matter of how you weigh them up. I don’t have anything but instinct to justify the weighing I’ve done.
Totally fair! I think part of my reasoning here relies on the difference between “sentence I think is longer than necessary for the purposes of sentencing” (which I would not necessarily classify as an “evil” in the common English usage of that term) and an “unhinged” result. I would not support a consistent sentence if it were unhinged (or even a close call), and I would generally split the difference in various proportions if a sentence fell in places between those two points.
It’s a little hard to define the bounds of “unhinged,” but I think it might be vaguely like “no reasonable person could consider this sentence to have not been unjustly harsh.” Here, even apart from the frame of reference of US sentencing norms, I cannot say that any reasonable person would find throwing the book at SBF here to have been unjustly harsh in light of the extreme harm and culpability.
Yeah, sorry, when I said “unhinged” I meant “the US penal system is in general unhinged”, not “this ruling in particular is unhinged”. I also used “evil” as an illustrative / poetic example of something which I’d rather be inconsistent than consistent, and implied more than I intended that the sentencing judge was actually doing evil in this case.
It’s possible that I’m looking at how the system treats e.g. poor people and racial minorities, where I think it’s much more blatantly unreasonable, and transplanting that judgement into cases where it’s less merited. 25 years is still a pretty long time though, and I wouldn’t personally push for longer. (I would, however, support a lifetime ban from company directorships and C-suite executives and similar.)
I understood the gist in context as ~ “using US sentencing outcomes as a partial framework, or giving weight to consistency when many sentences are excessive or even manifestly so, poses significant problems.” And I think that is a valid point.
Your last sentence raises another possible difference in how to approach the question: My reactions to how long he should serve are bounded by the options available under US law. I didn’t check, but I think the maximum term of supervised release (the means of imposing certain post-sentence restrictions) is only a few years here. And there is no discretionary parole in the federal system, so I can only go off of SBF’s lack of remorse (which requires acknowledgement of wrongdoing, not just mistakes-were-made) in assessing his future dangerousness. It’s possible I would go down somewhat if I could maintain a tight leash on post-sentence conduct in exchange.
Finally, I think it’s appropriate to consider a few other practical realities. It is practically essential to give defendants an incentive to plead guilty when they are actually guilty; that is commonly thought of as 25-33%. Likewise, we have to further punish defendants who tamper with witnesses and perjure themselves. The US system detains way too many people pre-trial, and if we’re going to fix that then I think the additional sanction for abusing pre-trial release has to be meaningful. So I have almost a doubling of the sentence here compared to a version of SBF who pled guilty, didn’t tamper, and didn’t perjure. So to me, saying 25 years was enough here implies ~12.5 would be enough for that version of SBF, with about 9-10 years estimated actual incarceration.
You make a lot of good points. I think there’s a lot of practical realities of an effective system here that I didn’t confront, and honestly, I’m probably better off leaving that stuff to those who know more about it, like you :)
he got (and many white-collar criminals get) significantly less than his culpability level and harm caused would predict.
What do you think is the correct level of punishment for white collar crimes based on harm? If we only look at first-order effects [1], even stealing 1B is just really bad, consequentially. Like if we use a simple VSL framework it’s equivalent to ~100 murders.
But of course, this is very much not how the justice system currently operates, so overall I’m pretty confused.
[1] And not looking at second order effects of his crimes e.g. creating negative reputation for EA which is probably negative (Though prosecutors might disagree), or creating negative reputation for crypto and political donations which is probably positive.
As far as what they predict, 40-50 years as explained in the government’s sentencing memo.
As far as what the impact should be—I would have to write a book on that. To start with, I see multiple, related harm-related measures here:
The amount of expected harm the offender knew or should have known about (this is the culpability-flavored measure);
The actual expected value of harm (this is more of a general deterrence-flavored measure); and
The actual harm (more of a retribution-flavored measure).
I also don’t see a unified measure of harm in economic-loss cases, as the harm associated with stealing $1,000 from a working class person is substantially higher than the harm of stealing it from me. Targeting vulnerable victims also gets you an enhancement for other reasons (e.g., it suggests a more extensive lack of moral compass that makes me value incapacitation more as a sentencing goal).
But most fundamentally, both harm and culpability go into the mix, filtered through the standard purposes of sentencing, to produce a sentence that I think is sufficient, but not greater than necessary, to accomplish those goals.
So I can tell you that the relationship between harm and sentence in fraud cases shouldn’t be −0-, both because there is little or no general deterrence against making your frauds bigger, and because there is some relationship between culpability and fraud size. It also shouldn’t be linear, both because this is impractical given the wide variance in harms, and because the degree of culpability does not ordinarily vary in a linear manner.
Most people in the sentencing realm think the federal sentencing guidelines increase the sentence too much based on loss amount (~ 25% uplift for each doubling of amount, plus some other uplifts tend to scale with size) and give too much weight to loss size in general. I agree with both of those views. Roughly and after considering a fuller measure of harm than aggregate financial loss, I might consider such an algorithm appropriate for weighing the harm factor in arriving at a sentence rather than for uplifting the sentence itself. And maybe I would increase the average sentence about 10% for each doubling, if all other factors were equal (low confidence)? But usually non-harm culpability factors increase with harm size, so the average increase in practice would probably be higher (low confidence).
As a bonus exercise: The differential impact between harm and non-harm culpability factors seems to vary quite a bit based on offense type. In drunk driving cases, actual harm seems to explain the bulk of the variance? No harm, first offense -- 2 days. Kill someone, no harm—perhaps 10 years. The non-harm culpability factors and even the expected-harm measures aren’t really that different in a lot of cases. Even here, the extra punishment for multiplied harm isn’t linear; you might get a 50% uplift for the second fatality, and a total of 85% more for three fatalities (low confidence)?
On the other hand, if you rob five convenience stores on five days before getting caught, you might get twice the sentence for robbing one (low confidence) -- even though the culpability scales here much more obviously with the number of offenses / amount of harm. One theory here is that getting caught and punished gives you the opportunity to learn your lesson; if you rob again after going through that then the argument for long-term incapacitation is stronger, etc.
It feels like you’re arguing for a higher-than-necessary level of harm and suffering in the world, just because a high level of suffering already exists in this context? I can’t see an argument with this structure working anywhere else (and believe me, I think Sam should be punished).
For some purposes of sentencing, social context affects what is reasonably “necessary” and what I would count as “harm.” Society is the other participant in the sentencing exercise.
Suppose A thinks from first principles that five years was enough for SBF. In the context of the American system, that sentence would convey something about offense severity—and it would drastically understate it where a street dealer in illegal drugs might get the same sentence. So because of the background severity of sanctioning, the five-year sentence is inadequate to convey the appropriate degree of societal condemnation. If you think the social educative function of the criminal process is important, that’s a problem. A sentence that tells everyone in the US that what SBF did isn’t really that big a deal on the grand scheme of criminal activity would be quite harmful.
The social meaning also matters to the victims. Seeing the person who harmed you receive a penalty in line with their crimes mitigates the suffering of many victims; seeing them get off easy imposes additional suffering on others. And those victims are experiencing the sentence through the lens of what other offenders in their society receive.
But the main harm I had in mind is the harm to the norm that all are equal before the law and will be treated equally. Now, if I hand out below-norm sentences to all offenders, maybe that particular harm isn’t present. But those kinds of breaks are not equally distributed; we know that the system is harder on people of color, poor people, and other disadvantaged groups. Giving SBF too much of a break from what others with comparable culpability and harm receive gives the perception that the privileged are more favored before the law than the commoners. Or maybe the reality—some people have argued that the court should consider SBF’s loss of reputation and loss of massive earning capacity, which sounds a whole lot to me like taking his privilege into account. Avoiding that harm is more important to me than whether SBF spends a few extra years in prison.
(I should clarify that I think a sentence of more than 25 years was necessary and appropriate here on/near first principles, mainly for reasons described in the quick take linked to my comment. I should note that this case hits a lot of the aggravating factors I tend to weigh heavily, and I see no significant mitigating factors here at all.)
I think this changed my mind, but at least for me I was more persuaded by your first point. I momentarily forgot that I really believe that white collar crime should have huge deterrent punishments that take little regard to the personal circumstances of the defendant; that ultimately a large punishment for Sam would proportionally create much less future harm. And that that’s not inconsistent with desiring & working toward an end to the hellish US prison system.
I think this is a fair compromise between what the prosecutors wanted and what the defense wanted; I don’t have an opinion on what’s the “correct” level of punishment for this type of crime. My guess is that if I did a first-principles analysis his crime is either the type of thing that gets ~5 years or something that gets life imprisonment without parole, but I’m not confident and also I don’t see much value in forming my own independent impression on optimal deterrence theory, given that it’s not decision-relevant to me at all.
SBF did terrible acts from many different moral viewpoints, including that of consequentialism. In addition to those he directly harmed, he harmed the EA movement.
However, from review of what I have read, it seems as if he acted from a sincere desire to better the world and did so to the best of his (quite poor) judgment. Thus, to me, his punishment is a tragedy, though a necessary one. From a matter of ultimate culpability, I don’t know if I would judge him more harshly than the vast majority of people in the developed world: those having the capability to save or dramatically better the lives of people in the developing world, but decline, or those who thoughtlessly contribute to the torture of animals through their participation in the animal product economy.
I wish him comfort and hope that he can find a wiser path forward with the remainder of his life.
Although none of us can peer into SBF’s heart directly, I think a conclusion that he acted from mixed motives is better supported by the evidence. It would take a lot to convince me that someone who was throwing money around like SBF on extravagances (or a $16.4MM house for his parents) was not motivated at least in considerable part by non-benevolent desires.
If one thinks he viewed luxuries bought as part of his fraudulent enterprise as ultimately altruistic because what makes SBF productive --> makes FTX richer --> makes the world better through philanthropy, then we have a framework under which it is impossible to categorize his motives because any behavior can be recast as ultimately motivated by altruism. The base rate of fraudsters being motivated by personal gain is very high, so—unless there’s a clear way to verify lack of personal-gain motivation—I think assuming its absence is doubtful.
I disagree, but it obviously depends what exactly we’re discussing.
Was his judgment for not coming clean when things were only starting to get bad compromised by not wanting to lose his influence, money, and reputation? Probably!
However, do I think he made some of his most consequential decisions to a significant degree because he thought he could get nice things for himself that way? I actually don’t think so!
Making big decisions for reasons other than impact would ruin his score in the game. It looks to me like his primary drive was optimizing his life for impact like a video game, trying to score the most points. Making a big life decision for self-oriented reasons would be forfeiting points he could otherwise have gotten, which would feel deeply unsatisfying for someone who’s always thinking about how to get the highest score.
Also, $16 million is peanuts for someone who’s personally worth $2-10 billion, which SBF probably “”somewhat reasonably”″ thought he was worth when things were going well (or when he optimistically thought he could make them well again soon enough). It’s like what a couple hundred bucks is to someone with a net worth of $80,000 – not that significant! Imagine you’re making decisions for tens of millions every day. I can’t emphasize enough how unimportant it then becomes whether you pay 1 million or 16 million for your parents’ house (which you may or may not also plan to use for yourself or some of your coworkers/flatmates). Just like with offices or the building where he planned to sleep in, the price isn’t the relevant variable (what’s more relevant is, “Does it take up your time or people’s time with high opportunity costs? Is it close to the office where you work? Is it convenient?” Etc.)
I feel like,
if someone displays as much single-minded obsessiveness as SBF,
and his conversations with the people around him tell you he’s hyperfocused on impact (he wasn’t just obsessing about how to grow FTX for money, he was also constantly discussing ideas around grantmaking strategies or paying Trump to not run again for election),
and his close social circle is reinforcing the impact-orientedness,
and there’s a credible ideology behind it which generated many other examples of similarly impact-obsessed people (hopefully the vast majority of them without the recklessness and “no fear of bad consequences” part),
then I’d say the evidence points to being primarily impact-driven pretty much as strongly as it gets. Of course, no one has the motivations of a perfect utilitarian robot, but I think SBF was trying, so it doesn’t seem fair to call it “acted from mixed motives.”
Edit to add: We could discuss whether he was “unusually biased” for someone who’s trying to have a ton of impact – hopefully yes (otherwise EAs don’t stand a chance of getting impact right), but that’s still meaningfully different from “acting from mixed motives.” (Also, I’d say the most severe bias was probably a blindspot around risk taking and risks of things going badly, rather than e.g., being too much into a rich lifestyle or being too much into fame or whatever. FWIW, my best guess is that he also was too much into wanting power/influence; I’m just saying that it doesn’t seem like the primary reason things went so poorly.)
Edit2: Somewhat relevant comment I made a year ago about why I think SBF having dark triad traits is compatible with the above picture.
If you’re going to compare outflows to guess at motivation, I think it’s better to use actual charitable giving numbers as the comparator rather than perceived wealth. Doubtless underinclusive on both numbers, but IIRC the grants through FTXFF were ~$150MM and the outflows to his parents alone were ~$25MM (there was also ~$10MM in cash). That doesn’t suggest the self-serving cashflows were peanuts in either a relative sense or an absolute sense.
If you assumed based on past actions that ~ 90% of the money was going to end up donated to charity, then ~$1B of the loss found by the district court today was attributable to the 10% that wasn’t. Given SBF’s crimes, and the fairly brazen nature of his perjury at trial, I would credit his observed actions over what he said he’d do with all the money in the end.
That’s a good point. If there weren’t a convincing story for why more donations weren’t made or at least set up to be made soon, I’d say your point counts for quite a lot!
However, in this specific case, I feel like there are good reasons why I wouldn’t expect that many donations to be made right away:
FTX did have a hole in the bank! It’s interesting that donations were made at all given that they were strictly speaking insolvent. (Of course, he had to keep up appearances or pay for previously-made commitments, etc., so it’s not too surprising. I’m just saying it probably wouldn’t have been wise even for someone as risk tolerant and unconcerned-by-the-illegality-of-it as SBF to donate much more when there was still a hole in their wallet.) (Edit: Admittedly, this point also cuts against giving money to his parents.)
Longtermist grantmaking (which SBF thought was most impactful, I believe) already started to feel a bit crowded once FTX Future Fund came in, so it’s not actually that easy (or sensible) to deploy $100s of millions on short notice in that cause area.
The FTX Future Fund was still relatively new; it makes sense to scale up giving as you learn things and do various types of preparatory work for your grantmaking/”active grantmaking.”
“Donating now vs later” isn’t actually the most obvious of EA strategy questions, esp. if you think you’re greatly beating the market on investment returns, which he had been doing at least in his own mind.
SBF’s entire philosophy was about massive wins and game-changing ambitious stuff, so he probably thought of the donations he had already made or was making at the time as not that significant compared to what he was gonna do later, and probably had tentative ideas for long-term plans like “amass enough money to buy a chip company and then use that as leverage to get more AI safety work done at labs,” or something super ambitious like that, for which it felt worthwhile to keep investing in the growth of his FTX empire.
Lastly, and somewhat related to the previous bullet point of weird super-ambitious ideas, he did discuss the paying Trump thing and, at least according to Michael Lewis’s sources, he was entertaining the thought (though probably not super seriously because this was when they had a massive hole in the bank?) of paying a sum for it that would have been a lot bigger than the $150 million in charitable contributions you mentioned. (But sure, you can say that, since nothing happened there, it was only talk. We don’t know, but I find it plausible that he’d have liked the thought of being the guy who prevented Trump from running!)
Edit: If someone says “giving 10 million to his parents (and a house, but we’re not sure if the house had other uses) shows he must have had mixed motives,” I’d be like: Okay, yeah, it does look like this isn’t what a utilitarian robot would’ve gone for, but when I hear “mixed motives,” I think of something like 50-50 or at least 40-self-oriented, 60-altruism, whereas giving money to get your parents to retire early could also be compatible with something like 10-90 (or even 5-95), “focus on impact for the big decisions, but do something nice for your parents once you’ve made it big.”
SBF likely had mixed motives, in that there was likely at least some degree to which he acted in order to further his own well-being or with partiality toward the well-being of certain entities (such as his parents). The reasoning that you mentioned above (privileging your own interests instrumentally rather than terminally such that you as an agent can perform better) is a fraught manner of thinking with extremely high risk for motivated reasoning. However, I think that it is one that serious altruists need to engage with in good faith. To not do so would imply giving until one’s welfare was at the global poverty line, which would probably impair one too much as an agent. Of course, I’m not saying he was engaged in good faith regarding this instrumental privileging argument, but I cannot preclude the possibility.
Regardless, I have been persuaded by everything that I have seen that a significant part of SBF’s motivations were to help advance a world of higher well-being. Of course, from a deontological perspective he did wrong by his dishonest and fraudulent actions. From a consequentialist perspective, the downside risks had such incalculable costs that it was terrible as well.
But the sincere desire of his to make the world a better place makes me sympathetic of him in a way that I probably would not be with similarly sentenced other convicts. Given a deterministic or random world, I understand that all convicts are victims too. But I cannot help but feel more for one who was led to their crime by a sincere desire to better the world, than say, to kill their spouse in a fit of rage, or advance themselves financially without any such altruistic motivation.
This is just not true if you read about the case, he obviously knew he was improperly taking user funds and tells all sorts of incoherent lies to explain it, and it’s really disappointing to see so many EAs continue to believe he was well-intentioned. You can quibble about the length of sentencing, but he broke the law, and he was correctly punished for it.
Please note that my previous post took the following positions:
1. That SBF did terrible acts that harmed people.
2. That it was necessary that he be punished. To the extent that it wasn’t implied by the previous comment, I clarify that what he did was illegal (EDIT: which would involve a finding of culpable mental states that would imply that his wrongdoing was no innocent or negligent mistake).
3. The post doesn’t even take a position as to whether the 25 years is an appropriate sentence.
All of the preceding is consistent with the proposition that he also acted with the intention of doing what he could to better the world. Like others have commented, his punishment is necessary for general deterrence purposes. However, his genuine altruistic motivations make the fact that he must be punished tragic.
All punishment is tragic, I guess, in that it would be a better world if we didn’t have to punish anyone. I guess I just don’t think the fact that SBF on some level “believed” in EA (whatever that means, and if that is even true) - despite not acting in accordance with the principles of EA—is a reason that his punishment is more tragic than anyone else’s
As others have observed, it’s been taboo to publicly speculate on SBF’s innocence since long before his conviction, including—or perhaps especially—in EA spaces. So please excuse the pseudonym, but...
I think this sentence is a tragedy. This isn’t the place to get into my general issues with the United States incarceration system, but even putting that to one side, I believe there’s much more to this story than meets the eye and it would be nice to see more EAs seriously grappling with the arguments in his defense.
I think people who have followed this unusually closely should be encouraged to argue for what they think is right if they have a strong take, but I just don’t think this theory is likely. An innocent person would be more likely to talk more freely about things/be less evasive and they’d probably have a better explanation of how it is that they could have missed an 8 billion hole in the bank. It’s suspicious if you need to make the same move (“he could’ve not seen this” or “he could have not looked closely at that”) multiple times to preserve the chance of innocence. Every time you use a not-too-likely excuse like this, your hypothesis takes a hit.
I don’t have a strong view on the sentence length. I think the main reason for a low sentence length (compared to the guidelines for the counts he was convicted of) is that I do believe SBF was strongly altruistically motivated, which is unusual for cases like that. I think the main reason for a long(-ish) sentence length is perjury and evasiveness.
A better explanation than what?
It wasn’t evasiveness, in my view.
Plus I’m not sure if the judge means much by “perjury” beyond “continues to plead not guilty” and this so-called evasiveness.
Pleading not guilty is not perjury. For starters, it isn’t a claim of innocence, merely a demand that the prosecutors put up or shut up. Second, perjury has a specific definition, which includes that a material, false statement was made under a legally valid oath or its equivalent.
SBF testified under oath in his own defense at trial. He was not obliged to testify, and the jury could not draw a negative inference from his decision not to. In fact, most US defendants do not testify.
If he made a material, false statement with knowledge of its falsity under oath at trial, then he committed perjury. Testifying under oath that you “don’t remember X” when you know you actually do is perjury (assuming that X is material, a rather low bar to meet).
Example (quoting the Government’s sentencing memo):
I agree that some of his behavior was just unproblematic “being very literal about answers.”
But the thing I mean by evasiveness was more stuff like:
Not remembering important things
Not giving answers that substantially clarified what happened
Saying “I don’t know which piece of evidence you’re referring to?” when he was several times asked simple questions on whether he did or didn’t do something. Note that this is the opposite of taking questions too literally; instead, it’s being deliberately obtuse to mask his refusal to ever disclose information that the prosecution didn’t already present, which is a central example of “being evasive.”
Regarding perjury, a lawyer (I believe it was an FTX lawyer?) testified that SBF asked him what excuses he could use to explain that the money is gone, then the lawyer suggested some pontential avenues but was like “they won’t work because it’s not actually legal/doesn’t hold up with the details” (paraphrased), and SBF nodded at that, but then used these excuses anyway talking to the media?! And then when asked about what he had said in the media, he didn’t take it back, which at that point constitutes perjury.
You have my respect for acknowledging that; it’s very rare that anyone does.
I really think people should cut him a lot more slack for this. On November 11th, new FTX management said in a public announcement that, “Sam Bankman-Fried has resigned his role as Chief Executive Officer and will remain to assist in an orderly transition”, while behind the scenes they were reporting him to authorities[1]. So he probably didn’t expect that he was about to be permanently shut off from access to internal company data to the greatest extent that new management could manage and therefore he may not have been secretly exporting all the relevant data to his personal storage as may have been prudent. (In fact, he’d been busy turning off autodelete “on any place I found it”.) It’s very hard to remember exactly how things played out when in the middle of a crisis you’re suddenly shut off from almost all the relevant data and it’s now a year or so later.
He also gave dozens of interviews. A very literal person, when asked, “Did you say xyz on this date?” is not going to think, “I expect they’re quoting me...but sometimes they’re just paraphrasing me in a misleading way...and obviously I don’t actually remember...but it’s going to look really bad if I say I don’t remember and then they put a tweet from me on the screen saying it...so I better say yes.”
He was also living off bread and peanut butter in one of the worst jails in the country with limited access to the internet, discovery material and his ADHD meds, he’d just gone through a year of the globe turning on him with the final nail in the coffin being his former girlfriend testifying extensively against him right in front of him, and, I expect, he was doing what he could on the stand to mask autistic mannerisms so that he wouldn’t come across so unlikeable and shady. This can’t have helped his ability to recall things.
I’d be interested to hear examples (genuinely). I can’t think of times when it seemed like this is what was happening, rather than him just “being autistic” (or not actually remembering, see above). In that link in my last comment it says, for example, “In their article Autism in the Criminal Justice System, Judge Kimberly Taylor and other experts urge criminal justice professionals to ‘[u]understand the need to repeat and rephrase questions” and note that “difficulty in maintaining eye contact or insistence on changing the subject of conversation to a topic of their choice—all typical diagnostic behaviors of a person with autism—can mislead an investigator, attorney, or judge.” Judy Mark—a UCLA Disability Studies faculty member who has an autistic son...acknowledges “answers that seem non-responsive’ and ‘times when [autistic individuals] might repeat themselves or become obsessed on an issue, something we call perseveration. They might become particularly focused on minute details that others find irrelevant, something we call hyperfocus.’”
Yes, an FTX lawyer[2]. Again, I’d be interested to hear more details. It sounds to me like they didn’t realize the nature of the mistake they’d made until the weeks leading up to the run on the exchange and I expect they thought that, given their ignorance at the time, they had not committed fraud. But I also expect that now they were facing a liquidity crisis—and didn’t think anyone would believe them that it was an accident—they were keen to know if it would have been fraud even if they had known exactly what they were doing. I can’t think of a time he lied in the media on the basis of this (or, ever).
In fact, Nishad appears to slip up at one point on this: “However, not all elements of the prosecution narrative line up neatly. Singh said he left the crucial June meeting still thinking things were OK and did not realise customer funds were being raided until September” (Financial Times).
I’m getting almost all of this from the first website I linked to, by the way.
“On November 9, 2022 . . . S&C attorneys in our Criminal Defense & Investigations Group, in consultation with Mr. Miller, reported the concern to federal authorities, including the United States Attorney’s Office for the Southern District of New York [USAO], the Securities and Exchange Commission and the Commodity Futures Trading Commission.” (from this paper)
Under an immunity grant. FTX lawyers were permitted to testify for the prosecution but not the defense, I believe.
I don’t want to spend too much time on this so won’t answer to all points, but I wanted to point you to some examples for this bit about evasiveness by saying things like, “I don’t know what this is referring to”:
See the transcript here: the word “referring” occurs 30 times and at least a couple of those times strike me as the weasel-like suspicious behavior of someone whose approach to answering questions is “never admit to anything unless you learn that they already have the evidence.” So, he always answers first with “not sure/don’t know what you refer to/don’t remember” and only admits to things when shown evidence.
This behavior is strikingly abnormal and different from how a person who doesn’t have anything to hide would behave.
(Edit – and again, it seems to me like it’s different from autistic literal-mindedness! Literally answering the question would mean to comb your memory and answer without regard for what the prosecution is referring to. It would also include saying confidently “no” if you’re sure you never said something.)
Someone trustworthy would answer questions immediately, sometimes admitting to things that the prosecution may not already know.
Some examples:
[...]
[...]
To me, the focus on “What this is referring to” is illuminating because it shows how SBF is laser-focused on what the prosecution has on him. What’s strikingly absent is a desire to try hard at remembering so he can tell as much of the truth as possible.
Thanks for going to the trouble of sourcing these quotes.
I mean, in the “practice” cross-examination, I think he was frequently rebuked for guessing at what the prosecutor meant or sharing information beyond a short, binary, direct answer to the question the prosecutor (thought she) was asking. For example, “A. And by “go negative,” you’re talking about negative in a particular coin or negative net asset value? Q. Just have a negative balance, Mr. Bankman-Fried. A. Sorry. I— … THE COURT: … I’ve gotten beyond my tether here. … part of the problem is that the witness has what I’ll simply call an interesting way of responding to questions for the moment. Q. Mr. Bankman-Fried, in May of 2022, were you aware that account ID 9 @AlamedaResearch.com could have an overall negative value? A. I am giving you my best guess at answering the question. Q. I’m not asking for a guess. I’m asking what you understood at the time. A. I am going to answer what I think the question you are asking is, but I apologize if I’m answering the wrong question.” I think the precise question she intends to ask here is extremely important, but she won’t specify (or doesn’t understand that she needs to.) It’s a very complicated case and the details matter. If Alameda’s trading accounts collectively had a net liability of billions to FTX for an extended period with no uncorrelated collateral, then they “borrowed” billions here in the way ~everyone thinks they did. But if one subaccount was temporarily negative $3 billion in a particular coin while another subaccount was temporarily positive $4 billion in another coin and it’s mostly for market-making purposes, then the sense in which Alameda was “borrowing” $3 billion via its customer accounts is extremely different.
And then there are plenty of reports of how when SBF gave extra information that wasn’t asked for, that was surely “evading questions, trying to pour forth verbiage to distract Sassoon from what she’d asked.” (from here)
Perhaps we’re understanding the term “referring” differently—I hear it as “I don’t understand exactly what you’re asking me (and if I guess wrong you might make me look like a liar)” while you might be hearing it as “I don’t know which of the many pieces of supporting evidence have prompted you to ask this question and I don’t want to reveal more than you already know about.”
Taking the first example of “referring to” in your comment, we have:
I’m picturing his thought process as something like, “I think I used the word ‘safe’ multiple times in public forums in reference to FTX U.S. If that is what she’s referring to then my answer is ‘Probably.’ But maybe she’s referring to that tweet someone else wrote using the FTX Twitter account that said FTX International was safer than Binance. And I can’t think of any times I’ve actually described FTX International with the word ‘safe’ and I don’t think I would have said that, so what do I say? If I say ‘Yes’ then I’m giving the impression that I said something bad that I don’t think I said, but if I say ‘No’ then she might bring up that tweet and make me look like a liar.”
I think that’s how my mind would respond anyway, given the way other interactions had gone. For example, “Q. Now do you remember telling Zeke Faux in early 2022 that Alameda played by the same rules as other traders? A. Not in that wording, no. Q. So you don’t recall that. A. No. Q. Do you recall telling him that in other wording? A. I recall saying that Alameda wasn’t front running other customers, that its trading access was like other customers.” If SBF had given a straight “no”—because he did not in fact make the claim that the prosecutor is heavily implying he did—and then the prosecutor had shown the jury a quote from Faux paraphrasing SBF that was taken out of context, would that have been perjury? I would definitely have been wary of giving yes/no answers to imprecise or misleading questions.
Who is ever sure they never said something? I’m not sure why you and I disagree here. I feel like the “normal” way to answer questions about things you may or may not have said a year or so ago is to say “yes” when you’re >50% sure you said it and “no” otherwise. Or, perhaps when testifying in your own defense, to just say “yes” every time you get the sense the prosecutor is about to pull out some written record of you saying it, even when that’s technically lying because you don’t actually remember, because it’s not worth the risk. But I think literal-minded people will feel very uncomfortable having only those two options and would rather say “I don’t recall”, especially if in squeezing their 25% confidence into a “no” only to get proven otherwise would constitute perjury.
I don’t know which of the lines above are the 2+ suspicious ones, but none of them strike me as such.
I’d note that Judge Kaplan invited the defense to seek a delay in the trial date if it felt that SBF being remanded to jail prejudiced their ability to put on a defense. They did not do so (likely in an attempt to create a longshot appellate issue, in my opinion). He also had access to a lot of discovery material for many months before being remanded.
There’s also a bit of tension between SBF’s claim that he didn’t have access to information needed to prepare his defense, and that the government dumped too much information on his defense team (at least 7.7 million pages, if I read correctly). Now, I don’t know what was in the Brady material, but I’d want specifics before concluding that important information was unavailable to the defense.
That’s interesting. Do you know when that invitation was (I can’t find it)? If it was close to the trial date, fair enough. If it was at the point of being remanded, it looks like he ended up experiencing a lot of problems that weren’t particularly foreseeable.
I can fact check if you’re curious, but just going by memory for now:
In SBF’s “practice” testimony that the judge requested, someone made a comment about how very convenient it was that he wasn’t able to produce the document retention policy that he kept referring to in an effort to defend himself. He replied that he’d requested it from FTX many times.
There was at least one complaint in the run up to the trial that a huge amount of discovery material had been made available to the defense at such a late date, and SBF was having so many problems with Internet access, that this should hardly count as an opportunity to review the information.
There’s also a huge difference between “information the prosecution wants to use against the defendant” and “information the defendant wants to use to defend himself”. I think the basic situation was that SBF didn’t have a realistic opportunity to look at the former before the trial and that he never had access to the latter (except for information that was inextricably part of evidence that prosecutors had judged served their own ends more than SBF’s overall.) I don’t think there’s a tension between millions of pages being used against you and having minimal access to data that might exonerate you.
(Docket order of Sept 12, 2024, citing transcript of August 30). He was remanded on August 11, so it was pretty soon after that. Moreover, any competent defense attorney would know that they could seek a trial delay on this basis; they did not have to wait for Judge Kaplan to broach the possibility himself.
The district court addressed the late production argument in the same order:
(typos in docket text).
“[I]nformation the prosecution wants to use against the defendant” isn’t an accurate characterization of the government’s Brady obligations—see the Wikipedia article linked earlier. “[P]art of evidence that prosecutors had judged served their own ends more than SBF’s overall” is pretty much the polar opposite of what Brady requires. And I think it is rather unlikely that USAO/SDNY would play games with Brady and thereby risk what they saw as a slam-dunk conviction, especially where the defendant was represented by well-bankrolled counsel and had at least some inside knowledge of what documents might be out there.
Also, none of this seems tied to the specific perjury findings the Government asked the district court to consider—that “Bankman-Fried falsely testified that he did not know about Alameda spending customer deposits, that he did not know until late 2022 that FTX’s code features were being improperly used to take customer money, and that he thought in November 2022 that his tweets were accurate and not misleading.” (p. 31-32 here).
Finally, as to evasiveness / hair splitting / dodging more generally, I’d note that Judge Kaplan has been on the bench for thirty years and has plenty of exposure to defendants who have been remanded, witnesses and defendants who may have conditions that affect their ability to give testimony, and those in extremely complex cases with voluminous discovery. I see no reason to believe he was incapable of factoring those things in to the extent relevant, or unwilling to do so, and trust his findings of fact more than SBF’s self-serving narrative unless given a good reason to depart from that.
From a media report:
Thank you—I misread you, I apologize, I thought you’d said the judge offered a trial delay. (Although it’s still somewhat interesting that Judge Kaplan reminded the defense that they could submit a request for a delay and he would consider it.)
Clearly the Court did not find this persuasive, but I still want to share some sense of the difficulties SBF was facing: at “the latest opportunity the defense had to observe whether the plan for access to the internet-enabled laptop would work as promised,” due to a number of factors, SBF “spent an entire day getting access to a single document. That is one day he could not spend working at the MDC or conferring with his lawyers.” (from here). Perhaps Brady only requires that access is given to SBF’s lawyers (who were not all “well-bankrolled” by the way—Mills at least worked pro bono), but I still think these kind of things are worth mentioning in discussions of the extent to which SBF’s testimony was evasiveness vs. genuinely not recalling.
Also, even if “the recently produced documents” were mostly “from the defendant’s own Google accounts,” he had at least 2.5 million pages in his Google accounts, so whether they were technically “accessible” is not really the point. (Assuming he did have access—SBF reported that FTX cut off his access to his personal LinkedIn account, for example.)
But I do think I veered into talking about how FTX would only provide data to the prosecution not the defense, when my original comment that started this thread was just the words “discovery material”, which may have confused things.
I think Judge Kaplan’s apparent dislike of SBF has been widely reported and I assume we’ll have to agree to disagree on whether that biased any of his judgments in this case.
Yeah, we’ll have to agree to disagree on that.
I don’t have videotape of Judge Kaplan, of course, but some behavior that may come across as “apparent dislike” could actually reflect sound courtroom management techniques. For instance, sternly lecturing an defendant who does things like use a VPN while on tight electronic lockdown is a restrained use of the district court’s power, authority, and responsibility to manage the proceedings before it. It can be an attempt to prod an errant defendant back in line without bringing out the heavier-handed powers (although Judge Kaplan did end up remanding SBF after the second tampering episode, on which he was affirmed by the Second Circuit).
To be perjury, false, material testimony must be under a legally valid oath or its equivalent. Lying to the media isn’t perjury; generally it is not a crime at all.
What I meant with by “he didn’t take it back” is a situation as follows:
The prosecution asks him if he made certain claims in the media. SBF says “yes” or “it appears that way” or whatever. The prosecution at some other point in the trial (maybe days earlier, maybe afterwards) asks some specific details about how FTX accounts were structured and how money was moved that contradicts what SBF said in the media. At some third point in the trial, they ask him if he deliberately lied to the media/gave false accounts about how things worked, and he said no. (Or, instead of asking him “did you tell the truth to the media?,” maybe they just asked him the same question the media asked him, and SBF stuck to his guns to avoid admitting that he mislead the media – similar outcome because he’s now saying the false thing in the trial setting.)
(I’m not actually sure this exact thing happened – it’s been a while since the trial. But I’d guess there were cases like that where you can look at different responses he gave to various questions and find that they’re in tension with things established by the prosecution from other witnesses. That’s probably also what the judge meant when he said that he found three instances of perjury? But I’m flagging that I haven’t looked at Kaplan’s elaborations on those three points. It’s more that I trust that Kaplan is likely to come to the right interpretations on those counts because when I followed the trial, it also felt to me like SBF was, at various points, getting trapped/caught in a web of questions, previous answers, and others’ testimony. In other words, I’m pretty sure I thought at various points “this is probably perjury established, if we believe this other witness?” – but I didn’t write those down, so I don’t remember the exact web of testimonies that would’ve conclusively trapped him.)
He got off easy, in my opinion. I wrote earlier here about the need for general deterrence, as well as enhancement for certain aggravating factors like witness tampering and perjury.
For non-U.S. readers, the sentence may seem pretty harsh; the actual time to serve may end up roughly equal to the median UK single murderer who did not bring a weapon to the scene and whose conduct did not involve any other statutory aggravating factor. But, compared to what the US criminal justice system regularly hands out for various levels of moral culpability and harm caused, he got (and many white-collar criminals get) significantly less than his culpability level and harm caused would predict. I think “rich people crime” getting a significantly less stern response than ordinary crime is damaging to the rule of law and the social contract in this country.
I also feel little pity for him, at least relative to the median US convicted individual. A large number of people who receive significant prison sentences experienced great childhood trauma, suffer from significant mental illness that contributed to their offenses, and had relatively few good options in life. SBF had great privilege and could choose among an extremely broad range of lawful and attractive paths in life. I don’t think anyone has suggested that any mental illness or autism was contributory to his offense.
So in this case, my desire for consistency and inter-defendant fairness trumps the broader concerns I have about the US system being too punitive.
While I see what you’re saying here, I prefer evil to be done inconsistently rather than consistently, and every time someone merely gets what they deserve instead of what some unhinged penal system (whether in the US or elsewhere) thinks they deserve seems like a good thing to me.
(I don’t personally have an opinion on what SBF actually deserves.)
To clarify, you would sacrifice consistency to achieve a more just result in an individual case, right?
But if there could be consistently applied, just, results, this would be the ideal result...
I don’t understand the disagree votes if I am understanding correctly.
I can’t speak for the disagrees (of which I was not one), but I was envisioning something like this:
You are one of ten trial judges in country X, which gives a lot of deference to trial judges on sentencing. Your nine colleagues apply a level of punitiveness that you think is excessive; they would hand out 10 years’ imprisonment for a crime that you—if not considering the broader community practices—would find warrants five years. Although citizens of county X have a range of opinions, the idea of sentencing for 10 years seems not inconsistent with the median voter’s views. The other judges are set in their ways and have life tenure, so you are unable to affect the sentences they hand down in any way. Do you:
(a) sentence to five years, because you think it is the appropriate sentence based on your own judgment;
(b) impose a ten-year sentence that you find excessive, because it prevents the injustice of unequal treatment based on the arbitrary spin of the assignment wheel; or
(c) split the difference, imposing somewhere between five and ten years, accepting both that you will find the sentence too high and that there is an unwarranted disparity, but limiting the extent to which either goal is compromised.
I would be somewhere in camp (c), while Ben sounds like he may be closer to camp (a). I imagine many people in camp (b) would disagree-vote Ben, while people in camp (c) might agree, disagree, or not vote.
yes, that’s right.
I can think of grounds to disagree, though. Say for example you were able to disproportionately protect e.g. white people from being prosecuted for jaywalking. I think jaywalking shouldn’t be illegal, so in a sense any person you protect from prosecution is a win. But there would be indirect effects to a racially unfair punishment, e.g. deepening resentment and disillusionment, enabling and encouraging racists in other aspects of their beliefs and actions. So even though there would be less direct harm, there might be more indirect harm.
I think the indirect harms are at work in this case too, and it’s just a matter of how you weigh them up. I don’t have anything but instinct to justify the weighing I’ve done.
Totally fair! I think part of my reasoning here relies on the difference between “sentence I think is longer than necessary for the purposes of sentencing” (which I would not necessarily classify as an “evil” in the common English usage of that term) and an “unhinged” result. I would not support a consistent sentence if it were unhinged (or even a close call), and I would generally split the difference in various proportions if a sentence fell in places between those two points.
It’s a little hard to define the bounds of “unhinged,” but I think it might be vaguely like “no reasonable person could consider this sentence to have not been unjustly harsh.” Here, even apart from the frame of reference of US sentencing norms, I cannot say that any reasonable person would find throwing the book at SBF here to have been unjustly harsh in light of the extreme harm and culpability.
Yeah, sorry, when I said “unhinged” I meant “the US penal system is in general unhinged”, not “this ruling in particular is unhinged”. I also used “evil” as an illustrative / poetic example of something which I’d rather be inconsistent than consistent, and implied more than I intended that the sentencing judge was actually doing evil in this case.
It’s possible that I’m looking at how the system treats e.g. poor people and racial minorities, where I think it’s much more blatantly unreasonable, and transplanting that judgement into cases where it’s less merited. 25 years is still a pretty long time though, and I wouldn’t personally push for longer. (I would, however, support a lifetime ban from company directorships and C-suite executives and similar.)
I understood the gist in context as ~ “using US sentencing outcomes as a partial framework, or giving weight to consistency when many sentences are excessive or even manifestly so, poses significant problems.” And I think that is a valid point.
Your last sentence raises another possible difference in how to approach the question: My reactions to how long he should serve are bounded by the options available under US law. I didn’t check, but I think the maximum term of supervised release (the means of imposing certain post-sentence restrictions) is only a few years here. And there is no discretionary parole in the federal system, so I can only go off of SBF’s lack of remorse (which requires acknowledgement of wrongdoing, not just mistakes-were-made) in assessing his future dangerousness. It’s possible I would go down somewhat if I could maintain a tight leash on post-sentence conduct in exchange.
Finally, I think it’s appropriate to consider a few other practical realities. It is practically essential to give defendants an incentive to plead guilty when they are actually guilty; that is commonly thought of as 25-33%. Likewise, we have to further punish defendants who tamper with witnesses and perjure themselves. The US system detains way too many people pre-trial, and if we’re going to fix that then I think the additional sanction for abusing pre-trial release has to be meaningful. So I have almost a doubling of the sentence here compared to a version of SBF who pled guilty, didn’t tamper, and didn’t perjure. So to me, saying 25 years was enough here implies ~12.5 would be enough for that version of SBF, with about 9-10 years estimated actual incarceration.
You make a lot of good points. I think there’s a lot of practical realities of an effective system here that I didn’t confront, and honestly, I’m probably better off leaving that stuff to those who know more about it, like you :)
What do you think is the correct level of punishment for white collar crimes based on harm? If we only look at first-order effects [1], even stealing 1B is just really bad, consequentially. Like if we use a simple VSL framework it’s equivalent to ~100 murders.
But of course, this is very much not how the justice system currently operates, so overall I’m pretty confused.
[1] And not looking at second order effects of his crimes e.g. creating negative reputation for EA which is probably negative (Though prosecutors might disagree), or creating negative reputation for crypto and political donations which is probably positive.
As far as what they predict, 40-50 years as explained in the government’s sentencing memo.
As far as what the impact should be—I would have to write a book on that. To start with, I see multiple, related harm-related measures here:
The amount of expected harm the offender knew or should have known about (this is the culpability-flavored measure);
The actual expected value of harm (this is more of a general deterrence-flavored measure); and
The actual harm (more of a retribution-flavored measure).
I also don’t see a unified measure of harm in economic-loss cases, as the harm associated with stealing $1,000 from a working class person is substantially higher than the harm of stealing it from me. Targeting vulnerable victims also gets you an enhancement for other reasons (e.g., it suggests a more extensive lack of moral compass that makes me value incapacitation more as a sentencing goal).
But most fundamentally, both harm and culpability go into the mix, filtered through the standard purposes of sentencing, to produce a sentence that I think is sufficient, but not greater than necessary, to accomplish those goals.
So I can tell you that the relationship between harm and sentence in fraud cases shouldn’t be −0-, both because there is little or no general deterrence against making your frauds bigger, and because there is some relationship between culpability and fraud size. It also shouldn’t be linear, both because this is impractical given the wide variance in harms, and because the degree of culpability does not ordinarily vary in a linear manner.
Most people in the sentencing realm think the federal sentencing guidelines increase the sentence too much based on loss amount (~ 25% uplift for each doubling of amount, plus some other uplifts tend to scale with size) and give too much weight to loss size in general. I agree with both of those views. Roughly and after considering a fuller measure of harm than aggregate financial loss, I might consider such an algorithm appropriate for weighing the harm factor in arriving at a sentence rather than for uplifting the sentence itself. And maybe I would increase the average sentence about 10% for each doubling, if all other factors were equal (low confidence)? But usually non-harm culpability factors increase with harm size, so the average increase in practice would probably be higher (low confidence).
As a bonus exercise: The differential impact between harm and non-harm culpability factors seems to vary quite a bit based on offense type. In drunk driving cases, actual harm seems to explain the bulk of the variance? No harm, first offense -- 2 days. Kill someone, no harm—perhaps 10 years. The non-harm culpability factors and even the expected-harm measures aren’t really that different in a lot of cases. Even here, the extra punishment for multiplied harm isn’t linear; you might get a 50% uplift for the second fatality, and a total of 85% more for three fatalities (low confidence)?
On the other hand, if you rob five convenience stores on five days before getting caught, you might get twice the sentence for robbing one (low confidence) -- even though the culpability scales here much more obviously with the number of offenses / amount of harm. One theory here is that getting caught and punished gives you the opportunity to learn your lesson; if you rob again after going through that then the argument for long-term incapacitation is stronger, etc.
It feels like you’re arguing for a higher-than-necessary level of harm and suffering in the world, just because a high level of suffering already exists in this context? I can’t see an argument with this structure working anywhere else (and believe me, I think Sam should be punished).
For some purposes of sentencing, social context affects what is reasonably “necessary” and what I would count as “harm.” Society is the other participant in the sentencing exercise.
Suppose A thinks from first principles that five years was enough for SBF. In the context of the American system, that sentence would convey something about offense severity—and it would drastically understate it where a street dealer in illegal drugs might get the same sentence. So because of the background severity of sanctioning, the five-year sentence is inadequate to convey the appropriate degree of societal condemnation. If you think the social educative function of the criminal process is important, that’s a problem. A sentence that tells everyone in the US that what SBF did isn’t really that big a deal on the grand scheme of criminal activity would be quite harmful.
The social meaning also matters to the victims. Seeing the person who harmed you receive a penalty in line with their crimes mitigates the suffering of many victims; seeing them get off easy imposes additional suffering on others. And those victims are experiencing the sentence through the lens of what other offenders in their society receive.
But the main harm I had in mind is the harm to the norm that all are equal before the law and will be treated equally. Now, if I hand out below-norm sentences to all offenders, maybe that particular harm isn’t present. But those kinds of breaks are not equally distributed; we know that the system is harder on people of color, poor people, and other disadvantaged groups. Giving SBF too much of a break from what others with comparable culpability and harm receive gives the perception that the privileged are more favored before the law than the commoners. Or maybe the reality—some people have argued that the court should consider SBF’s loss of reputation and loss of massive earning capacity, which sounds a whole lot to me like taking his privilege into account. Avoiding that harm is more important to me than whether SBF spends a few extra years in prison.
(I should clarify that I think a sentence of more than 25 years was necessary and appropriate here on/near first principles, mainly for reasons described in the quick take linked to my comment. I should note that this case hits a lot of the aggravating factors I tend to weigh heavily, and I see no significant mitigating factors here at all.)
I think this changed my mind, but at least for me I was more persuaded by your first point. I momentarily forgot that I really believe that white collar crime should have huge deterrent punishments that take little regard to the personal circumstances of the defendant; that ultimately a large punishment for Sam would proportionally create much less future harm. And that that’s not inconsistent with desiring & working toward an end to the hellish US prison system.
I think this is a fair compromise between what the prosecutors wanted and what the defense wanted; I don’t have an opinion on what’s the “correct” level of punishment for this type of crime. My guess is that if I did a first-principles analysis his crime is either the type of thing that gets ~5 years or something that gets life imprisonment without parole, but I’m not confident and also I don’t see much value in forming my own independent impression on optimal deterrence theory, given that it’s not decision-relevant to me at all.
Nitpick: I think this is a linkpost not a repost since this is the first time it has been shared on the forum.
Correct. Wasn’t aware of the difference between the two terms. Title changed.