Pablo

Karma: 3,294

«If we are the only rational beings in the Universe, as some recent evidence suggests, it matters even more whether we shall have descendants or successors during the billions of years in which that would be possible. Some of our successors might live lives and create worlds that, though failing to justify past suffering, would have given us all, including those who suffered most, reasons to be glad that the Universe exists.» — Derek Parfit

Pablo 8 Mar 2021 14:13 UTC
5 points
on: Care and demandingness
I have nothing substantive to say, but wanted to note that I thought this was a really good post, and that I’m surprised it got little engagement.

Pablo 7 Mar 2021 23:45 UTC
2 points
in reply to: AnonymousEAForumAccount’s comment on: Our plans for hosting an EA wiki on the Forum
Thanks! Responded there.

Pablo 7 Mar 2021 23:43 UTC
8 points
in reply to: AnonymousEAForumAccount’s comment on: Our plans for hosting an EA wiki on the Forum
Hi Anonymous,

Thank you for your thoughtful and extended feedback. I appreciate the time you have taken to raise a number of valid concerns. I will just respond to a few of your points, since much of what you say is in reply to Aaron’s previous comment, and I don’t want to interfere with that conversation.

I think the worry that the Wiki may fail due to insufficient contributions is very real. As you note, none of the previous attempts to build something like what we are trying to accomplish here have succeeded. And it appears that this is a common phenomenon with general efforts to create specialist wikis. Forecasting is one of my hobbies, and I’m well aware that the base rates aren’t in our favor.

This was is fact my primary concern back when I was considering this project for a grant application. The reason I eventually decided to go ahead—besides feeling that I had a somewhat higher shot at success than my predecessors based on my experience editing Wikipedia and the insight this experience gave me about my capacity to feel motivated long-term by a project of this nature—was that I thought I could gain more information by just trying things out for a few months. The money costs for EA Grants were relatively modest, as were the time costs for me: at the time I didn’t have any other project I felt excited about, and I don’t think I would have spent those months very productively otherwise.

Fortunately, the experiment was a success: by the time the grant was over, I had not only produced more content than I had promised, but had discovered that I found writing these articles a more enjoyable experience than I had anticipated. Since then, I have continued to work on the project, and all the direct evidence indicates that lack of motivation will not be a serious impediment. This may still be insufficient to warrant an update from the naive prior to the point that I feel super confident that I will either continue to work on this project full-time for at least the next five years or find a suitable replacement, but if I had to guess, I’d estimate the chances of this happening at something like 65%.

How sensitive are your worries to scenarios in which the main paid content-writer fails to stay motivated, relative to scenarios in which the project fails because of insufficient volunteer effort? I’m inclined to believe that as long as there is someone whose full-time job is to write content for the Wiki (whether it’s me or someone else), in combination with all the additional work that Aaron and the technical team are devoting to it, enough progress will probably occur to sustain growth over time and attract volunteer contributors. I’m modestly confident in this, but I’m much more confident in that it makes sense to (again) test the hypothesis experimentally, by trying to make the Wiki happen and see how excited people feel about it after a period of a year or so.

Pablo 6 Mar 2021 16:00 UTC
2 points
on: List of EA-aligned email newsletters (did I miss any?)
The Anti-Apocalyptus Newsletter:

Each week I send you five links about some of the most important challenges of our time such as climate change, weapons of mass destruction, emerging technologies, mass causes of death and great power wars.

Pablo 6 Mar 2021 13:21 UTC
23 points
in reply to: MichaelA’s comment
on:
Error
The value NIL is not of type (VECTOR CHARACTER)

Pablo 6 Mar 2021 11:39 UTC
14 points
in reply to: Max_Daniel’s comment on: Our plans for hosting an EA wiki on the Forum

“hinge of history” or “patient philanthropy”

Those concepts definitely fail to meet Wikipedia’s notability standards, and I think they are a good example of why those standards are inadequate for the EA Wiki.

Pablo 6 Mar 2021 11:18 UTC
18 points
in reply to: AnonymousEAForumAccount’s comment on: Our plans for hosting an EA wiki on the Forum

My guess is that’s ~90% feature, 10% bug. I think most of the value of an EA Wiki would come from content that’s both EA relevant and notable by Wikipedia’s standards, and that the EA Wiki content most likely to go stale would be that which didn’t meet external notability standards.

How familiar are you with Wikipedia? I ask because I’ve been an editor for 17 years (though only active at certain periods) and my sense is that Wikipedia’s notability standards would be a very poor criterion to judge whether the EA Wiki should have an entry on a given topic. Such standards are in line with a “deletionist” editing philosophy for which I don’t see a plausible justification (see Gwern’s In Defense of Inclusionism for discussion) [I now see that Aaron had already linked to this essay]. Furthermore, even if those standards were justified for Wikipedia, I don’t think they would be for a specialized wiki, which is meant to be of interest to a much narrower audience and whose criteria for inclusion should reflect this specialist focus.

More importantly, if you want to get EA ideas into the mainstream, at some point you’re going to have to convince people outside EA that those ideas are notable. Wikipedia seems as good a place to do so as any, since it has established procedures for assessing new content and the payoff for success is getting EA ideas included in the world’s most accessible repository of knowledge.

I’m not sure I’d describe our goals as getting EA ideas into the mainstream; I think there’s some reasonable disagreement within the community concerning how much and how fast we want EA to grow. In any case, I do not agree that the most effective way of convincing outsiders that EA ideas are notable is to focus our energies on writing content directly on Wikipedia rather than on a separate wiki. Before embarking on this project, I tried that approach for a few months and felt pretty disappointed with the outcome. It was in part on the basis of that experience that I concluded this project should be done as a separate wiki.

To be clear: I think contributing to Wikipedia is extremely valuable—indeed, at the current margin I think one of the most impactful things a competent and dedicated EA could do is to work full time on improving the quality and coverage of EA content on Wikipedia, and I encourage anyone who thinks they may be a good fit for this to seriously consider this option (I’d also be happy to share my experience as an editor, so feel free to contact me privately to schedule a chat). My disagreement with you concerns not the value of contributing to Wikipedia, but the claim that Wikipedia is the best place currently to host the project of creating a comprehensive EA encyclopedia.

Pablo 3 Mar 2021 10:47 UTC
3 points
in reply to: Pablo’s comment
on:
Error
The value NIL is not of type (VECTOR CHARACTER)

Pablo 3 Mar 2021 10:37 UTC

3 points

on:

Error

The value
  NIL
is not of type
  (VECTOR CHARACTER)

Pablo 21 Feb 2021 11:35 UTC
2 points
on: UK to host human challenge trials for Covid-19 vaccines
Update from Vox’s Future Perfect: UK challenge trial will infect volunteers with Covid-19 to test vaccines.

a human challenge trial… is set to kick off in the UK within a month, now that an ethics board has given it the necessary approval. Scientists have run such trials in the past for influenza, malaria, typhoid, dengue, and cholera, but this will be the world’s first for Covid-19.

Only healthy volunteers aged 18 to 30 will be allowed to participate. The challenge trial is not meant to replace conventional clinical trials that are still ongoing for the different vaccines in development — the different types of trials will proceed along parallel tracks.

Many young people clamored to participate in the challenge trial. The advocacy group 1DaySooner gathered the names of nearly 39,000 people in 166 countries who said they’d be interested in volunteering.

The UK trial will begin with an initial phase that isn’t about testing vaccines at all; it’s about answering a preliminary question that lays the groundwork for testing vaccines later. The question is: How much of the virus does someone need to be exposed to in order to launch an infection?

Scientists will expose a small group of unvaccinated volunteers (somewhere between 50 and 90 people) to a very low dose of the virus, gradually increasing it until they find “the Goldilocks dose” — enough to infect most participants, but not enough to make them very ill.

the philosopher and MacArthur “genius grant” winner Rebecca Goldstein said signing onto 1DaySooner’s letter in support of challenge trials was “almost a no-brainer” for her.

“I have more faith both in people’s ability to understand the consequences of things, and in some people’s incredible sense of responsibility to others,” she told me last October. “Maybe it’s hard to imagine that there really are altruistic people, and we think, ‘Oh, they just don’t understand that harm could come to them.’ But I do believe there are such people.”

Pablo 19 Feb 2021 18:45 UTC
2 points
in reply to: Cullen_OKeefe’s comment
on:
Error
The value NIL is not of type (VECTOR CHARACTER)

Pablo 19 Feb 2021 11:50 UTC

2 points

on:

Error

The value
  NIL
is not of type
  (VECTOR CHARACTER)

Pablo 18 Feb 2021 12:30 UTC
22 points
in reply to: Cullen_OKeefe’s comment on: Charges against BitMEX and cofounders

(3) is a bit more complex, but would also be solved with anonymization.

Anonymization would probably solve (3), but would, unfortunately, likely create PR risks of its own. Lawrence Lessig made a similar argument a while ago:

Everyone seems to treat it as if the anonymity and secrecy around Epstein’s gift are a measure of some kind of moral failing. I see it as exactly the opposite. IF you are going to take type 3 money, then you should only take it anonymously.

Unfortunately, from what I can remember, public response to this argument was overwhelmingly negative, and The New York times (yes, that newspaper) published a story whose headline portrayed Lessig in a very bad light (Lessig subsequently filed a defamation lawsuit against the Times, which he withdrew after the headline was amended four months later). I personally would not have anticipated such a response, since the argument seems pretty reasonable to me, and I wonder if EAs as a whole may be apt to underestimate certain PR risks simply because they rely on their own subjective sense of the merits of the relevant arguments to predict how the broader public and the media will react to them.

Pablo 17 Feb 2021 12:27 UTC
9 points
in reply to: Linch’s comment on: Some global catastrophic risk estimates
The reasoning in the comment you quoted is actually not very persuasive, because it’s virtually certain that the user will be dead by 2100, Metaculus won’t exist by then, or MIPs will have ceased to be valuable to them. Even the slightest concern for accuracy should trump the minuscule expected benefit from pursuing this alleged “optimal strategy”. (Though I guess some would derive great pleasure from being able to truly say “I predicted that humanity had a 99% chance of surviving the century 80 years ago and, low and behold, here we are, alive and kicking!”).

Unfortunately, for questions with a shorter time horizon, that kind of argument may have some force. I feel ambivalent about discussing these issues, since I’m not sure how to balance the benefit of alerting others to the potential biases in Metaculus against the cost of exacerbating those biases, either by drawing attention to this strategy among predictors who hadn’t considered it, or by creating the impression that other predictors are using it and thereby eroding the social norm to predict honestly. I guess one can try to emphasize that, at least with questions whose answers have social value, adopting the MIP-maximizing strategy when it is in conflict with accuracy should be seen as a form of defection and those who do it should feel bad about it.

Pablo 29 Jan 2021 12:10 UTC
4 points
in reply to: MichaelA’s comment on: Notes on Schelling’s “Strategy of Conflict” (1960)
As an avid Anki user, I actually prefer a deck over a review or a summary, other things equal. Most people don’t use Anki or other spaced-repetition software, though, so they are unlikely to share this preference.

Incidentally, could you make the deck available as an apkg file?

Pablo 28 Jan 2021 12:07 UTC
5 points
in reply to: Carl_Shulman’s comment on: The Upper Limit of Value

The simulation hypothesis is a plausible way for our physics models to be quite wrong about the world in which the simulation is conducted, and further there would be reason to think simulations would be disproportionately conducted under physical laws that are especially conducive to abundant computation

The main reason for taking the simulation hypothesis seriously is the simulation argument, but that argument needs to assume that our physical models are broadly correct about reality itself and not just the “physics” of the simulation. Otherwise, there would be no warrant for drawing inferences from simulated sense data about the behavior of agents in reality, including whether these agents will choose to run ancestor simulations.

Pablo 20 Jan 2021 12:04 UTC
6 points
on: Hilary Greaves: The collectivist critique of the EA movement

I couldn’t find a video or recording of the actual talk from the Conference on the Ethics of Giving. If you happen to know of one, please share it in the comments!

She gave a closely related talk at EAG London 2017.

Pablo 19 Jan 2021 14:56 UTC

2 points

on:

Error

The value
  NIL
is not of type
  (VECTOR CHARACTER)

Pablo 17 Jan 2021 10:24 UTC
8 points
on: Privacy as a Blind Spot: Are There Long Term Harms in Using Facebook, Google, Slack etc.?
Kudos for offering testable predictions. Unfortunately, the links do not appear to be working. Also, I think the admins can embed the Elicit links so that users can predict without leaving the EA Forum, though I’m not sure. Maybe an admin could clarify.

Pablo 5 Jan 2021 16:09 UTC
14 points
on: Julia Galef and Angus Deaton: podcast discussion of RCT issues (excerpts)
Angus Deaton:

There’s never been an experiment on aspirin.

Noah Smith:

If I go to PubMed and search for “aspirin randomized controlled trial”, I get 7,586 results. There are reportedly 700 to 1000 clinical trials conducted on aspirin every year. There were also experiments involved in the invention of aspirin; people knew that salicylic acid helped with headaches, but extracting and buffering the chemical were both non-trivial tasks.