Tobias Häberli

Karma: 2,240

Tobias Häberli 12 Jun 2026 14:56 UTC
4 points
0 ∶ 0
in reply to: david_reinstein’s comment on: david_reinstein’s Shortform
I found the ‘founder deployment by end-2026’ the hardest to set. It comes a bit as a surprise at the end, as I was already taking into account some considerations before, and the descriptions seem to do as well (e.g. “assets after lockups, taxes, sale timing”, and “execution delays”).

Tobias Häberli 12 Jun 2026 7:45 UTC
6 points
1 ∶ 0
on: Building Ideas-First EA
I’m really excited about you & Sarah starting this! :)
I wouldn’t invest community building efforts in different settings until I was confident this ecosystem was thoroughly supported
It’s pretty notable that you’re only focusing on North American universities. Do you think this subset of universities is your comparative advantage? Or do you think that most community builders should primarily focus on that subset?

If it’s the latter, I’d be interested in hearing your thoughts on why efforts into other top universities (Oxford, Cambridge, Imperial, UCL, ETHZ, NUS, TUM, Melbourne, etc.) should be lower priority.

Tobias Häberli 4 Jun 2026 7:53 UTC
1 point
1 ∶ 2
in reply to: Benevolent_Rain’s comment on: Why we’re launching the Frontier Biodefense Fellowship
I generally think it’s fine to downvote/disagree without explaining. But if someone wants to share thoughts, we’d definitely be interested (also feel free to DM).

Tobias Häberli 1 Jun 2026 14:04 UTC
−6 points
0 ∶ 5
in reply to: qamaster’s comment on: A Revolutionary Method for Liberating Civilization from the Production of Luxury Goods
I think you should be much more confused & curious here than you seem to be.

Tobias Häberli 31 May 2026 13:50 UTC
−5 points
0 ∶ 4
on: A Revolutionary Method for Liberating Civilization from the Production of Luxury Goods
In the UK, the government allocates about $500 million a year to cancer research. At the same time, private individuals can invest an equivalent amount ($500 million) into purchasing a single football club. Which investment holds greater value for humanity?
In the process of the football club changing hands for $500 million, what resources are destroyed?

Tobias Häberli 21 May 2026 7:25 UTC
2 points
1 ∶ 1
in reply to: Tobias Häberli’s comment on: Jenba’s Quick takes
Junior EAs getting advice that helps them grow: “I’m getting reard on”

Tobias Häberli 21 May 2026 7:20 UTC
5 points
0 ∶ 1
in reply to: Matt Beard’s comment on: Jenba’s Quick takes
So you’re saying he reard you?

Tobias Häberli 19 May 2026 15:19 UTC
9 points
1 ∶ 1
on: Why are longtermists so much less focused on human extinction these days?
My sense is that there’s been a significant shift in how much longtermists prioritise non-extinction risks over the last few years. A decade ago people who were trying to ensure the flourishing of the universe trillions of years from now were very often focused on avoiding events that would kill all humans.
I agree that this shift has happened over the past 2 years. But I think 2 to 3 years ago EAs were unusually focused on extinction compared to a decade ago. I remember more discussions back then around positive visions for the longterm.
For what it’s worth, we just announced our first Frontier Biodefense Fellowship at Pivotal, which is more singularly focused on avoiding extinction than most projects within AI safety (including our AI safety fellowship). Obviously the team has a range of motivation to work on Biodefense, but for me weak longtermist arguments are quite central.

Tobias Häberli 12 May 2026 8:50 UTC
8 points
0 ∶ 1
in reply to: Rohin Shah’s comment on: Reflections on Anthropic and EA
What’s your explanation for why they attack EAs rather than, say, the AI ethics crowd?
I think there are a few plausible reasons that don’t require “undemocratic power-seeking” as the primary explanation:
- EAs have been more motivated to & competent at gaining influence through legitimate/standard means
- EA policy ideas enjoyed better reception by policymakers, were more aligned with their interests. The opponents might have stronger disagreement with AI ethics ideas, but see them as much less likely to have influence.
- EAs compete more directly on their territory. The AI ethics crowd has sufficiently different values and assumptions that they’re less of a threat locally. (Kind of a ‘narcissism of small differences’.)
I expect that if your ideas are resonating with policymakers and people are getting appointed to relevant roles because they’re competent, bad faith opponents will target you roughly the same as if you’d been pulling strings behind the scenes in dubious ways.
Why was SB 1047 so controversial, while other much more onerous AI bills (esp for “little tech”) were barely discussed?
Maybe I’m missing something. SB 1047 seemed like a relatively transparent action, that followed the democratic process. Is your point that undemocratic power-seeking actions prior/unrelated to SB 1047 likely explains the stronger opposition to SB 1047?

Tobias Häberli 29 Apr 2026 9:10 UTC
9 points
0 ∶ 0
on: The Saturation View
I haven’t done philosophy in a while, might be missing something, but wanted to highlight what I think is the strongest objections to the view^[1] in a way that may be more salient than the framing in section 6. It’s probably a reason why many might prefer a total view.
To be clear, I do think the Saturation View improves on other non-total views I know of, and I appreciate that they flag some of its hard-to-stomach implications. But I still think the post understates how bad the separability issue is. So here are two short points:
Non-separability is really bad.
The core problem is that facts about/experiences of wholly unaffected people can change the value of the affected person’s experiences. If there are already sufficiently many people elsewhere with sufficiently similar experiences, then an additional person having an extremely deep, meaningful, happy life adds near-zero marginal value. That seems very hard to accept.
And for negative experiences the implication is potentially even less intuitive. An additional torturous experience can add almost no marginal disvalue if enough sufficiently similar torture already exists. They discuss this under the “cheap suffering” problem & call it the strongest argument against the view, but I think it is worth emphasizing just how unintuitive of a conclusion this is. From the victim’s perspective, the torture is not any less bad because other similar torture already occurred. But the saturation view says that, from the point of view of population value, their torturous experience would matter hardyl at all.
ETA: Relatedly, the view assigns value to our experiences depending on empirically inaccessible facts. Whether sufficiently distant aliens have sufficiently similar experiences is something we probably can’t know, but it would radically change how our actions matter. That seems strange.
I don’t think the ‘tameness’ of the view recovers that much?
My understanding is that the Saturation View does better because violations of separability are localized. Ancient Egyptians or distant aliens only affect the marginal value of new lives if their experiences are sufficiently similar. So in many “normal situations”, the view behaves roughly separably.
But the separability worry still holds with sufficiently large numbers. If enough sufficiently similar unaffected lives exist elsewhere, they can radically change the marginal value of what we do here.
And population ethics is full of large-number objections. The Repugnant Conclusion itself gets its core intuitive force from considering sufficiently enormous populations, and is also not a “normal situation.” So if the Saturation View is partly motivated by avoiding the very bad large-number implications of total views, then its own large-number implications seem fair game too.
1. ^
  the authors agree with this, afaict

Tobias Häberli 2 Apr 2026 6:14 UTC
1 point
0 ∶ 1
in reply to: MichaelDickens’s comment on: An unexplained annual spike in false claims on the EA Forum
I’m sad to announce that I’m leaving academia.
I’m looking forward to working on AI safety.

Tobias Häberli 1 Apr 2026 11:49 UTC
6 points
0 ∶ 1
in reply to: NickLaing’s comment on: An unexplained annual spike in false claims on the EA Forum
Good point, they might know. Does anyone know a neurotypical? Or a friend of a neurotypical that I could reach out to?

Tobias Häberli 1 Apr 2026 10:41 UTC
5 points
0 ∶ 0
in reply to: Bella’s comment on: Tapping into EA’s most neglected market
Attacking people’s outfit choices is a unique low for the EA forum!

Tobias Häberli 1 Apr 2026 10:39 UTC
23 points
7 ∶ 0
on: We’re growing: CEA is increasing its font size
THE EA FORUM TEAM COULD REALLY HELP HERE BY ADDING A ‘HEADING 0’ AND ‘HEADING −1’ FORMATTING OPTION.

Tobias Häberli 30 Mar 2026 11:45 UTC
11 points
4 ∶ 0
on: What is integral altruism?
Thanks for writing this!
You’re describing integral altruism as broader than EA, but if I understand you correctly, it’s also narrower in many ways. Some examples:
Letting go of the need to control everything and transcending the frame that we are in conflict with the natural unfolding of the universe. This also means emphasising collective action over individual heroism.
–> Effective altruism doesn’t take a position on whether we are in conflict with the natural unfolding of the universe. EAs emphasise collective actions vs. individual heroism to various degrees.
take radical uncertainty seriously
–> EAs already do this to various degrees. If integral altruists take this really seriously, they are a subset of EAs in this regard.
altruism grounded in truth rather than being driven by guilt or pride
–> EA doesn’t say where your altruistic motivation should be grounded in. All of the reasons you list are considered viable (although people of course disagree to what degree they are conducive/to be encouraged).

Some of the things you describe (especially the ‘different ways of knowing’) seem to sit more outside of what is common within EA. There it seems more like integral altruism actually is broader.

Overall I’m not completely sure whether integral altruism is a way of doing effective altruism differently, or a competing (though often overlapping) world view.

Tobias Häberli 26 Mar 2026 18:30 UTC
2 points
0 ∶ 0
in reply to: Davidmanheim’s comment on: $1 billion is not enough; OpenAI Foundation must start spending tens of billions each year
Good points, thank you!
They have incredibly short AGI timelines, so per their own views, they can’t afford to move slowly. If they are giving less than 5% of assets after they already claim AGI, that’s a huge failure.
Do we know whether this is true for the OAF board?^[1] Sam Altman is on it, and he definitely believes something along these lines but it’s less clear for the others. Here’s a ChatGPT and a Claude answer on this, which points towards the others being less bullish & concerned (but also a lack of information about what they believe). I expect there to be a range of views on timelines & transformativeness of AGI among the board members – which probably makes it more likely that their spending targets are compatible with the foundations mission.
1. ^
  Bret Taylor (Chair), Adam D’Angelo, Dr. Sue Desmond-Hellmann, Dr. Zico Kolter, Retired U.S. Army General Paul M. Nakasone, Adebayo Ogunlesi, Nicole Seligman, Sam Altman

Tobias Häberli 26 Mar 2026 11:13 UTC
11 points
2 ∶ 0
in reply to: Bella’s comment on: Bella’s Quick takes
It looks much nicer than the original imo. If I didn’t have context, I’d probably be confused though.
Why 80,000 hours? And what is the pie chart / watch face analogy about? On first glance I’m not sure whether it’s about career choice, time management, life balance, or some ‘5pm’ metaphor.
I looked at it in this order: (1) “80,000 hours”, (2) pie chart / watch face, trying to figure it out, (3) subtitle, (4) endorsement. But the subtitle and endorsement are doing most of the work of telling me what the book is actually about and whether it’s for me.

Maybe some of this is intended, to make people pick up the book and try to find answers. :)

Tobias Häberli 25 Mar 2026 17:29 UTC
7 points
2 ∶ 1
on: $1 billion is not enough; OpenAI Foundation must start spending tens of billions each year
I agree it would be bad if the OpenAI Foundation were still giving under 5% per year several years from now. But I don’t think ‘they should spend 5%+ in year one’ follows.
Directing billions well is really hard, especially for a new foundation. Coefficient Giving says it directed over $4 billion from 2014 to mid-2025, and that 2025 was the first year it directed more than $1 billion. Their ‘endowment’ is much smaller (~10x smaller?) than OAF’s but it still points towards allocating money well at that scale being genuinely hard. I wouldn’t call a new foundation planning to deploy $1 billion in its first year “conservative”.

What I’d most like to see is OAF committing to an aggressive, public ramp-up targets, maybe something like reaching 5% of assets by 2028.

Tobias Häberli 13 Mar 2026 14:42 UTC
8 points
2 ∶ 0
in reply to: quinn’s comment on: quinn’s Shortform
No, sorry. The diamond emoji (🔸) is specifically for people who donate 10% of their earnings.
But taking a 50% pay cut for altruistic reasons is incredibly based, so you should use the square emoji instead (🟧). It’s also larger, which seems fitting.

Tobias Häberli 4 Mar 2026 11:21 UTC
3 points
1 ∶ 0
in reply to: titotal’s comment on: Why isn’t anti-fascism a bigger topic at EAG events (or on this forum)?
Thanks, that’s useful. I mostly agree with you, and mistakenly read the second bullet point as saying “work that opposes fascism should come from all sides of the political spectrum”, which is something I agree with. I think the OP somewhat assumed that opposing fascism will look like ‘work with your local anti-fascist network’, but I expect much of it could look more like ‘militarising Europe’ (something the political left would typically oppose).

Tobias Häberli

THE EA FORUM TEAM COULD REALLY HELP HERE BY ADDING A ‘HEADING 0’ AND ‘HEADING −1’ FORMATTING OPTION.