SebK

• SebK 7 Sep 2021 17:44 UTC

  35 points

  0 ∶ 0

  in reply to: MikeJohnson’s comment on: A Primer on the Sym­me­try The­ory of Valence

  Hi Mike, I really enjoy your and Andrés’s work, including STV, and I have to say I’m disappointed by how the ideas are presented here, and entirely unsurprised at the reaction they’ve elicited.

  There’s a world of a difference between saying “nobody knows what valence is made out of, so we’re trying to see if we can find correlations with symmetries in imaging data” (weird but fascinating) and “There is an identity relationship between suffering and disharmony” (time cube). I know you’re not time cube man, because I’ve read lots of other QRI output over the years, but most folks here will lack that context. This topic is fringe enough that I’d expect everything to be extra-delicately phrased and very well seasoned with ifs and buts.

  Again, I’m a big fan of QRI’s mission, but I’d be worried about donating I if I got the sense that the organization viewed STV not as something to test, but as something to prove. Statistically speaking, it’s not likely that STV will turn out to be the correct mechanistic grand theory of valence, simply because it’s the first one (of hopefully many to come). I would like to know:

  1. When do you expect to be able to share the first set of empirical results, and what kinds of conclusions do you expect we will be able to draw from them, depending on how they turn out? Tiny studies with limited statistical power are ok; “oh it’s promising so far but we can’t share details” isn’t.

  2. I hope QRI’s fate isn’t tied to STV – if STV can’t be reconciled with the data, then what alternative ideas would you test next?

• SebK 24 Jan 2021 22:24 UTC

  29 points

  0 ∶ 0

  in reply to: sadiedec’s comment on: Why I’m con­cerned about Giv­ing Green

  Giving Green should recommend donating to a portfolio of promising policy change and activism organizations.

  As jackva points out, there is a thin line between effective advocacy for policy change (e.g. Clean Air Task Force) and the kind of activism that prevents conservative politicians from touching the climate file with a ten-foot pole, because their base sees it as a “leftist agenda” issue.

  Anecdotally, I have met with the staffers of several deep-red, lukewarmist/​denialist Republican senators to lobby for revenue-neutral carbon taxes (CCL is a great organization to volunteer with!). Surprisingly, they all privately agreed with the urgency of the climate crisis, but the only thing they really cared about during those meetings was evidence—like the Yale Climate Opinion Maps, hand-written letters from constituents, etc.—that supporting climate legislation wouldn’t get them primaried.

  In light of that experience, and the razor-thin margins held by Democrats, I can only agree with jackva that amplifying TSM/​XR/​Standing Rock/​etc. may well be counterproductive in some places. I’m glad Keystone XL won’t happen, but I would sleep better if clean tech and carbon pricing had killed its viability instead.

New study in Science im­plies that tree plant­ing is the cheap­est cli­mate change solution

[Link] Up­dated Draw­down now available, incl. 2020 Review

• SebK 8 Sep 2021 9:07 UTC

  20 points

  0 ∶ 0

  in reply to: MikeJohnson’s comment on: A Primer on the Sym­me­try The­ory of Valence

  I feel like your explanations are skipping a bunch of steps that would help folks understand where you’re coming from. FWIW, here’s how I make sense of STV:

  1. Neuroscience can tell us that some neurons light up when we eat chocolate, but it doesn’t tell us what it is about the delicious experience of chocolate that makes it so wonderful. “This is what sugar looks like” and “this is the location of the reward center” are great descriptions of parts of the process, but they don’t explain why certain patterns of neural activations feel a certain way.

  2. Everyone agrees that clearly, certain activation patterns do feel a certain way. Quite plausibly, this isn’t just a brain thing but more fundamental, and evolution simply recruited the relationship to build RL agents like ourselves. And yet, almost nobody has tried to figure out how exactly the patterns relate to the experiences. Of course, that’s because we struggle with both sides of the equation: on the neural side of the equals-sign, the data is incomplete and noisy; on the experience side, what do we even put to represent “delicious”?

  3. But we have to start somewhere, so we simplify. On the neural side, we look for the simplest kind of signature we can reliably detect in global-scale brain data: symmetries/​harmonies across space and time. On the experience side, we collapse everything onto a good/​bad axis: valence. Now that’s still a pretty vague hypothesis, but just barely solid enough that we can at least reason about and perhaps even test it.

  4. This seems very arbitrary! Well, collapsing qualia onto a single valence dimension (for now) is relatively uncontroversial, since there are at least a few things that everyone can agree feel fantastic or terrible. But why look for symmetry/​harmony/​resonance in the brain data, rather than other things like amplitude or spatial distribution? Here it’s worth explaining that you didn’t just pull that idea out of your … uhm, hat, but that experience suggests that all our senses – visual, spatial, temporal, auditory, etc. – are exquisitely attuned to certain kinds of symmetry. This may sound trivial, but the evidence from psychedelic research, intense meditation, psychotherapy etc. suggests that there’s something about it that goes much deeper than just “kaleidoscopes are pretty”. And also, that the hypothetical mathematical object representing one’s brain state is so high-dimensional that a huge class of neural activation patterns will have some kind of symmetry or another, leaving plenty of room for agreement with existing neuroscience. This is what most of your material is about, at least to my understanding.

  5. How compelling this feels (and just feels!) to investigate is something most readers won’t appreciate unless they’ve experienced altered states of consciousness themselves. This is worth acknowledging explicitly, but not condescendingly: “this may be more difficult to relate to if you haven’t tried psychedelics”, rather than “you wouldn’t understand if you’re at a lower developmental stage”.

  6. But also, none of this proves anything yet. People used to think that a fever was the disease, simply because it was the most obvious symptom, so perhaps back then it would have been an obvious leap to claim temperature were fundamental and causal to the qualia of sickness. It’s possible that the symmetry story will turn out to be a dud in the same way, even though it feels very appealing now and is certainly worth investigating.

• SebK 24 Apr 2021 0:18 UTC

  13 points

  0 ∶ 0

  on: [Linkpost] New Oxford Malaria Vac­cine Shows ~75% Effi­cacy in Ini­tial Trial with Infants

  Incredible news.

  I read Derek Lowe’s post about it earlier today, and it only says that they’re now going into Phase III. Knowing next to nothing about vaccine distribution or malaria, I wonder: if that 77% number holds up, what can we expect the next decade or two to look like, malaria-wise? Clearly 77% isn’t quite 100%; will people risk it and forgo bed nets and antimalarials? How likely is it that boosters will be required every few years? How much will this cost? Etc. etc. Anyone care to share their informed guesses at how this will go?

• SebK 10 Sep 2021 7:09 UTC

  9 points

  0 ∶ 0

  in reply to: Holly_Elmore’s comment on: A Primer on the Sym­me­try The­ory of Valence

  Thanks Holly! I’m not advocating for STV, I’m just an interested layperson who’s followed QRI’s work for some time and felt frustrated with everyone here furiously talking past one another.

  Is the claim that the symmetry is the qualia of valence? How would symmetries and resonance be exempt from the hard problem any more than neuronal activation?

  Yep – if I understand it correctly, the reasoning goes something like “there’s nothing obviously special about biological neurons as a physical substrate, so maybe consciousness is fundamental to the universe but only emerges when physical systems interact with each other/​themselves in particular ways”. IIT seems to have that flavour, and STV as well. I don’t know if it solves the hard problem per se, but I can see why a fundamental theory is more appealing than just a brain map of reward/​aversion “centers” and the like.

  Do you think it should be compelling based on a trip? Is that real evidence?

  I wanted to be careful, that’s why I tried to emphasize the word “feels” :P

  Trips are compelling evidence that the space of possible conscious experiences is vast and unspeakably weird, and that our “normal” consciousness is just what evolution optimized for to help us get through the day. And so in the endeavour of cataloguing, systematizing, and eventually trying to model qualia, I would trust someone who personally appreciates the vastness of this space, and who is rigorous and detailed about its weirdness.

  This is dangerous territory, not just epistemically but politically. Drunk and stoned people’s “deep insights” tend to be dumb nonsense, so why should we trust other druggies’ claims? Sadly, academic psychedelics researchers struggle with this public perception, and their solution is to publish only on clinical applications, as if the changes in consciousness were an embarrassing side effect rather than central to the experience. QRI are the only team I know of who don’t implicitly privilege sobriety and instead explicitly talk about the space of possible qualia. That is something I really appreciate.

  As for whether symmetries/​harmonies in the qualia experienced on trips are a compelling enough reason to look for symmetries/​harmonies in brain data – I really don’t know. But I do think gears-level models of qualia would be useful, and since neuroscience is mostly silent on the topic, symmetries are as good a place as any to start ¯\_ (ツ)_/​¯

• SebK 12 Jun 2019 3:35 UTC

  9 points

  0 ∶ 0

  on: New Re­port Claiming Un­der­state­ment of Ex­is­ten­tial Cli­mate Risk

  Thanks for posting this. I am grateful they published this report, and I hope that their explicit reframing in terms of existential risk will get the EA community’s attention.

  The EA standpoint so far has been “lots of money is already being thrown at climate change, it’s mostly a question of policy now”. And that’s true. Good ideas are out there: fee-and-dividend carbon pricing, Project Drawdown, etc.; all it takes is political will. Unfortunately, in my experience, many EAs take this to mean that climate change is an issue they can’t help with.

  It’s true that it is difficult finding out which policy approaches are effective and politically viable, and it’s also difficult to convince others that the climate needs urgent attention. Therefore, it seems to me that there is still much potential for EA organizations to give better advice to those of us who want to contribute not as donors or full-time researchers, but as citizens doing some “effective activism” in their spare time. If anyone can point me to work that has been or is being done in this direction, I would be grateful.

Ap­ply for AI Safety Camp Toronto 2020!

• SebK 10 Sep 2021 7:22 UTC

  5 points

  0 ∶ 0

  in reply to: Holly_Elmore’s comment on: A Primer on the Sym­me­try The­ory of Valence

  Again, not evidence for anything, but seizures can apparently be incredibly blissful, so it all depends. STV proponents would probably say that depending on the subnetworks involved and the particular synchronicities in the firing patterns, it could be a pleasant seizure or an unpleasant one …

• SebK 11 Aug 2020 4:18 UTC

  5 points

  0 ∶ 0

  on: The ex­treme cost-effec­tive­ness of cell-based meat R&D

  I really appreciate this work, but wonder about the magnitude of the uncertainty in your analysis. Would it be possible for you to convert your calculation into a Guesstimate sheet?

• SebK 25 May 2020 15:09 UTC

  3 points

  0 ∶ 0

  in reply to: UriKatz’s comment on: Cli­mate Change Is Ne­glected By EA

  That sounds right to me. (And Will, your drawbridge metaphor is wonderful.)

  My impression is that there already is some grumbling about EA being too elitist/​out-of-touch/​non-diverse/​arrogant/​navel-gazing/​etc., and discussions in the community about what can be done to fix that perception. Add to that Toby Ord’s realization (in his well-marketed book) that hey, perhaps climate change is a bigger x-risk (if indirectly) than he had previously thought, and I think we have fertile ground for posts like this one. EA’s attitude has already shifted once (away from earning-to-give); perhaps the next shift is an embrace of issues that are already in the public consciousness, if only to attract more diversity into the broader community.

  I’ve had smart and very morally-conscious friends laugh off the entirety of EA as “the paperclip people”, and others refer to Peter Singer as “that animal guy”. And I think that’s really sad, because they could be very valuable members of the community if we had been more conscious to avoid such alienation. Many STEM-type EAs think of PR considerations as distractions from the real issues, but that might mean leaving huge amounts of low-hanging utility fruit unpicked.

  Explicitly putting present-welfare and longtermism on equal footing seems like a good first step to me.

• SebK 25 May 2020 6:11 UTC

  3 points

  0 ∶ 0

  on: Cli­mate Change Is Ne­glected By EA

  What a fantastic post. Thank you! Your frustration resonates strongly with me. I think the dismissive attitude towards climate issues may well be an enormous waste of goodwill towards EA concepts.

  How many young/​wealthy people stumble upon 80k/​GiveWell/​etc. with heartfelt enthusiasm for solving climate, The Big Issue Of Our Time, only to be snubbed? How many of them could significantly improve their career/​giving plans if they received earnest help with climate-related cause prioritization, instead of ivory-tower lecturing about weirdo x-risks?

  Can’t we save that stuff for later? “If you like working on climate, you might also be interested in …”?

  This isn’t to say that EA’s marginal-impact priorities are wrong; I myself work mainly on AI safety right now. But a career in nuclear energy is still more useful than one in recyclable plastic straw R&D (or perhaps it isn’t?), and that’s worth researching and talking about.

  I’ve spent a good bit of time in the environmental movement and if anyone could use a heavy dose of rationality and numeracy, it’s climate activists. I consider drawdown.org a massive accomplishment and step in the right direction. It’s sometimes dismissed on this forum for being too narrow-minded, and that’s probably fair, but then what’s EA’s answer, besides a few uncertain charity recommendations? Where’s our GiveWell for climate?

  Given that people’s concern about climate change is only going up, I hope that this important conversation is here to stay. Thanks again for posting!

• SebK 5 Jul 2019 12:33 UTC

  2 points

  0 ∶ 0

  on: Wor­ld­wide de­cline of the en­to­mo­fauna: A re­view of its drivers

  For those concerned about wild animals, such a quick rate of decline could give some reassurance (in addition to the theoretical arguments) that wild insect populations will be small in the long-run.

  For those of us more active in other cause areas, could you clarify what you mean by this? Are you coming from an anti-natalist angle here, and is that the prevalent position in the wild animal community? What are the additional “theoretical arguments” for expecting small insect populations?