Posts

A tentative dialogue with a Friendly-boxed-super-AGI on brain uploads 2022-05-12T19:40:48.113Z
Ramiro P.'s Shortform 2021-09-23T15:09:12.170Z

Comments

Comment by Ramiro P. (ramiro-p) on "Liquidity" vs "solvency" in bank runs (and some notes on Silicon Valley Bank) · 2023-03-22T19:32:35.011Z · LW · GW

It turns that the truth is more bizarre. From Matt Levine's Money Stuff:

But today at the Wall Street Journal Hannah Miao, Gregory Zuckerman and Ben Eisen have the actual, horrifying explanation, which is that the Fed’s computers go to bed at 4 p.m. and you can’t wake them up until the next morning

The point I stressed before on government bonds was right: SVB could have borrowed against them. But it seems like I was wrong: it could have borrowed against Agency MBS, too.

Comment by Ramiro P. (ramiro-p) on "Liquidity" vs "solvency" in bank runs (and some notes on Silicon Valley Bank) · 2023-03-15T11:44:10.365Z · LW · GW

in a nutshell (I don't have the time to write a treatise on this, every word I write is 20w i could have read instead): I'm pretty confident (status: ~ .6) that if SVB had 90bi in gov bonds HTM instead of MBS and the like, it wouldn't have failed. You can say it would have had losses (especially in 2022), it would likely have been bought, but not failed. I know of no bank that has suffered a run because they had too much gov long term bonds (from the corresponding government, and unless the gov defaulted, ofc) in the last century (if you have an example, please enlighten me); not only there's a very liquid market for them, but, in the last century (perhaps since Badgehot) central banks will let banks convert them into money easily - because tax payers will suffer no losses. On the other hand, MBS (and other similar derivatives) may be linked to credit risks (even Agency MBS) and it's quite unsure how your liquidity line will work, and the market is not so liquid; and that's why they offer higher yields - which is why they dominated SVB's HTM. Thanks for the memory refreshing lecture on the crisis of 2008. But I still remember almost everything

Comment by Ramiro P. (ramiro-p) on "Liquidity" vs "solvency" in bank runs (and some notes on Silicon Valley Bank) · 2023-03-15T11:14:55.761Z · LW · GW

Because I work for a regulator and am not allowed to do that? Also, many investors won't have enough incentives to read beyond what other investors are reading... except if, as you mentioned, u work w shortselling And shortsellers did make money in this case. So in this sense, the system works... but when it happens to a bank, that's not so cool

Comment by Ramiro P. (ramiro-p) on "Liquidity" vs "solvency" in bank runs (and some notes on Silicon Valley Bank) · 2023-03-14T12:47:55.267Z · LW · GW

and the whole world will care a lot more about this term in bank reports, basically forever.

I'm usually astonished w how seldom investors and supervisors read the fine print in annual reports. I don't think "this time will be different". Unless GPT-like automated report-readers step in (or maybe precisely because humans will leave this boring details to machines), we'll see it happen again.

Btw, I just noticed that $9.3bi of these $15.2 are MBS - yeah, the same type of security associated w the great crisis of 2007-2008. And the HTM total more than U$90 bi, $72bi of which are MBS and CMBS - so dwarfing their investments in bonds, and their $17.6bi in AFS.

From the post above:

...something called Agency MBS (which you can just read as "government debt with extra steps")

I'm no expert in US markets, but I don't think that's true. For instance, if you try to get a repo w them, you'll probably need a larger hair-cut than w gov bonds.
 

if people had learned to read bank reports, I'd expect to read more comments on this, instead of the last three pieces I read that basically just said SVB had too much gov bonds.

EDIT: after googling "svb mbs htm," I found tons of surces commenting on this. So, my bad. And most of all, thanks for this post & for this comment, rossry. I believe you saved me at least 1h of googling - to have a better grasp of the situation.

Comment by Ramiro P. (ramiro-p) on Toni Kurz and the Insanity of Climbing Mountains · 2022-08-24T02:05:55.685Z · LW · GW

Well, I'm kinda sure climbers wouldn't like to brain-upload to a utopia simulation, so maybe there's some connection between this and AI alignment.
(Curiously, just read today Will MacAskill's WWOTF mentioning he used to climb buildings in Glasgow in his teens, until he was almost cut open by glass...)

Comment by Ramiro P. (ramiro-p) on A tentative dialogue with a Friendly-boxed-super-AGI on brain uploads · 2022-07-14T19:21:12.703Z · LW · GW

Thanks, I believe you are right. I really regret how much time and resources are wasted arguing over the extension / reference of a word.

I'd like to remark though that I was just trying to explain what I see as problematic in FiO. I wouldn't only say that its conclusion is suboptimal (and I believe it is bad, and many people would agree); I also think that, given what Celestia can do, they got lucky (though lucky is not the adequate word when it comes to narratives) it didn't end up in worse ways.

As I point out in a reply to shminux, I think it's hard to see how an AI can maximize B's preferences in an aligned way if B's preferences and beliefs are inconsistent (temporally or modally). If B actually regards sim-B as another self, then its sacrifice is required; I believe that people who bite this bullet will tend to agree that FiO ends in a good way, even though they dislike "the process".

Comment by Ramiro P. (ramiro-p) on A tentative dialogue with a Friendly-boxed-super-AGI on brain uploads · 2022-07-14T19:03:08.901Z · LW · GW

Possibly. I said this AGI is “safer and more aligned”, implying that it is a matter of degree – while I think most people would regard these properties as discrete: either you are aligned or unaligned. But then I can just replace it with “more likely to be regarded as safe, friendly and aligned”, and the argument remains the same. Moreover, my standard of comparison was Celest-IA, who convinces people to do brain uploading by creating a “race to the bottom” scenario (i.e., as more and more people go to the simulation, human extinction becomes more likely – until there’s nobody to object turning the Solar System into computronium), and adapts their simulated minds so they enjoy being ponnies; my AGI is way "weaker" than that.

I still think it’s not inappropriate to call my AGI “Friendly”, since its goals are defined by a consistent social welfare function; and it’s at least tempting to call it “safe”, as it is law-abiding and does obey explicit commands. Plus, it is strictly maximizing the utility of the agents it is interacting with according to their own utility function, inferred from their brain simulation - i.e., it doesn’t even require general interpersonal comparison of utility. I admit I did add a tip of a perverse sense of humor (e.g., the story of the neighbors), but that's pretty much irrelevant for the overall argument. 

But I guess arguing over semantics is beyond the point, right? I was targeting people who think one can “solve alignment” without “solving value”. Thus, I concede that, after reading the story, you and I can agree that the AGI is not aligned - and so could B, in hindsight; but it’s not clear to me how this AGI could have been aligned in the first place. I believe the interesting discussion to have here is why it ends up displaying unaligned behaviour.

I suspect the problem is that B has (temporally and modally) inconsistent preferences, such that, after the brain upload, the AI can consistently disregard the desire of original-B-in-the-present (even though it still obeys original-B’s explicit commands), because they conflict with simulated-B’s preferences (which weigh more, since sim-B can produce more utility with less resources) and past-B’s preferences (who freely opted for brain upload). As I mentioned above, one way to deflect my critique is to bite the bullet: like a friend of mine replied, one can just say that they would not want to survive in the real world after a brain upload – they can consistently say that it’d be a waste of resources. Another way to avoid the specific scenario in my story would be by avoiding brain simulation, or by not regarding simulations as equivalent to oneself, or, finally, by somehow becoming robust against evidential blackmail.

I don’t think that is an original point, and I now see I was sort of inspired by things I read from debates on coherent extrapolated volition long ago. But I think people still underestimate the idea that value is a hard problem: no one has a complete and consistent system of preferences and beliefs (except the Stoic Sages, who “are more rare than the Phoenix”), and it’s hard to see how we could extrapolate from the way we usually cope with that (e.g., through social norms and satisficing behavior) to AI alignment - as superintelligences can do way worse than Dutch books.

Comment by Ramiro P. (ramiro-p) on Lies, Damn Lies, and Fabricated Options · 2021-11-02T20:22:36.603Z · LW · GW

My concern, which I interpret as being TAG's point (but with different words), is that your example of water vs. XYZ is immediately traceable (at least for anyone who knows the philosophical discussion) to Putnam's Twin Earth experiment. The way you express your point suggests you disregard this thought experiment - which is surprising for someone acquainted with it, because Putnam (at least when he wrote the paper) would likely agree that a substance with the same basic chemical properties of water would be water. He actually aims to provide an argument for semantic externalism - i.e., the idea that the meaning of "water" (or of other natural species) is H20, its chemical nature, and not the apparent properties commonly used as criteria to discriminate it (that it's a tasteless liquid...). He's so pushing against a conventionalist view about semantics (and philosophy of language), thus it's not about physics or ontology.

Comment by Ramiro P. (ramiro-p) on Ramiro P.'s Shortform · 2021-09-23T15:09:12.625Z · LW · GW

LW is quoted (in a kind of flattering way) in Simon Dedeo's awesome piece on the last Nautil.us issue. For spoiler lovers:

[...] Rationality is my ticket out. The only reason I can trust you is that you seem rational enough to talk to. But now you’re telling me that rationality is just a layer on top of the System—it’s just as irrational as the people I’m trying to escape. I don’t know which is worse: being duped by someone else’s priors, or being a biological machine.

Teacher: Don’t go too far. You’re a smart kid—you can iterate faster than most. You can match patterns better. Evolution set you up well. You’ll get better at predicting the consequences of your actions, and better at adapting your environment to your will. Rationality is systematized winning.

Ian: It’s not winning I’m worried about. It’s my mind. Maybe it’s silly, maybe it’s a fetish, but I want to know the truth. It’s the principle of the thing. Wanting to know the truth got me this far, but now the only option you’ve given me is believing in something I can’t see. If I know it at all, it can’t be through rational, scientific calculation. There’s some kind of extra-rational process I have to engage in—but what’s beyond the edge of reason?

Teacher: Many things. Dreams, intuition, transcendence, love, ascending the ladder, repetition and the leap of faith, philosophy itself ...

Ian: ... delusion, fairy tales, fascism!

Teacher: Childhood’s end.

Comment by Ramiro P. (ramiro-p) on When Hindsight Isn't 20/20: Incentive Design With Imperfect Credit Allocation · 2020-11-21T00:01:32.735Z · LW · GW

Strict liability in tort law seems to be a pretty obvious example, doesn't it? I mean, I guess a lot of corporate law can be seen as "vicariously holding a group accountable"

Comment by Ramiro P. (ramiro-p) on Creating better infrastructure for controversial discourse · 2020-06-21T17:44:17.564Z · LW · GW

Kialo is totally underrated.

Comment by Ramiro P. (ramiro-p) on April Coronavirus Open Thread · 2020-04-10T15:48:08.455Z · LW · GW

I've seen news about this study, but no preprint. It'd be really helpful if we could get it.

Comment by Ramiro P. (ramiro-p) on What will happen to supply chains in the era of COVID-19? · 2020-04-06T01:25:08.669Z · LW · GW

I'm more concerned with developing countries, particularly if they depend on international trade for food. Also, their goods are mainly transported by truck drivers travelling long distances, who may opportunistically demand better working conditions. In Brazil, they threatened to stop working, thus reminding everyone of a strike in 2018 that caused supply problems in major cities. Fortunately, they reconsidered it after the government started a mediatic campaign to make them feel valued.

Comment by Ramiro P. (ramiro-p) on April Coronavirus Open Thread · 2020-04-01T13:30:32.192Z · LW · GW

Does anyone have any idea / info on what proportion of the infected cases are getting Covid19 inside hospitals? This seems to have been a real issue for previous coronavirus.

I'd say there might be a stark difference between countries / regions in this area. Italian health workers seem to have taken a heavy blow. Also, 79 deaths in Brazil (total: 200) came from only one Hospital chain/ health insurer, which focus on aging customers (so, yeah, maybe it's just selection bias?).

(Epistemic status: low, but I didin't find any research on that after 30min, so maybe the hypothesis deserves a bit more of attention?)

Comment by Ramiro P. (ramiro-p) on Preprint says R0=~5 (!) / infection fatality ratio=~0.1%. Thoughts? · 2020-03-24T12:57:27.642Z · LW · GW

I noticed CDC claims 9 deaths from Diamond Princess, but I didn't find support in their source. WHO is still counting 8 deaths. I guess you're right, but I'd appreciate if you could provide the source.

They write "at the time of testing." The study I cite followed up with what happened to patients.

I know that. If you follow this discussion up to the beginning, you'll see that all I'm claiming is that the number of documented cases has been affected by selective bias, because asymptomatic / pre-symptomatic etc. cases are unlikely to be diagnosed.

Finally, I believe we both agree the current IFR is underestimating the true death rate, because many patients are still fighting for their lives. Actually, the authors of the preprint are not complete morons and estimate the "time-delayed IFR" in 0.12% (which I agree is too low), and make the following remark to explain the higher mortality in Wuhan:

These findings indicate that the death risk in Wuhan is estimated to be much higher than those in other areas, which is likely explained by hospital-based transmission [32]. Indeed, past nosocomial outbreaks have been reported to elevate the CFR associated with MERS and SARS outbreaks, where inpatients affected by underlying disease or seniors infected in the hospital setting have raised the CFR to values as high as 20% for a MERS outbreak.

I'm not saying this study is right. I'm just saying that, unless someone points a methodological flaw, "their conclusion is too different" is not a reason to discard it.

Comment by Ramiro P. (ramiro-p) on Preprint says R0=~5 (!) / infection fatality ratio=~0.1%. Thoughts? · 2020-03-24T00:19:46.895Z · LW · GW

Maybe CDC screwed their data, but they say 46.5% of the Diamond Princess cases were asymptomatic when tested: https://www.cdc.gov/mmwr/volumes/69/wr/mm6912e3.htm?s_cid=mm6912e3_w

I believe this might be a confusion between asymptomatic and pre/mildly symptomatic - but whatever: the claim at stake is that there's a ton of undocumented cases out there, not that they're asymptomatic

Comment by Ramiro P. (ramiro-p) on Preprint says R0=~5 (!) / infection fatality ratio=~0.1%. Thoughts? · 2020-03-21T14:19:38.251Z · LW · GW

I'm sorry, I'm not sure if I understood the relevance of asymptomatic : symptomatic ratio here. I think what's at stake in this article is the ratio undocumented : documented cases; it'll include not only asymptomatic, pre-symptomatic or mildly symptomatic people, but people who got really sick but couldn't be tested until Hubei had largely improved their testing capabilities.

I do think a 50:1 rate is surprising, though not impossible.

If 50% of the cases in South Korea are asymptomatic and so don't get tested, their true death rate would be ~0.4-0.5%; if you add people who got sick before their testing capability was improved, etc., it may be lower. But again, I really prefer to be pessimistic in my death rates.

Comment by Ramiro P. (ramiro-p) on Preprint says R0=~5 (!) / infection fatality ratio=~0.1%. Thoughts? · 2020-03-20T21:31:52.611Z · LW · GW

True, but Diamond Princess is full of oldies, and, despite South Korea massive testing, there might be selection bias - I guess people would only get tested if they had some symptom or contact with other infected persons (perhaps you're referring a more specific study?). Notice that, if the science study claiming 86% of the cases in Wuhan were undocumented were right, this would already imply a fatality rate of about 0.6%, below South Korea estimates.

Yet, I agree the fatality rate is surprisingly low, and it's just a statistical model.

Comment by Ramiro P. (ramiro-p) on How much delay do you generally have between having a good new idea and sharing that idea publicly online? · 2020-03-17T22:02:10.144Z · LW · GW

My average is a week, I guess. I only share ideas I can't falsify or take out of my head; and I try them with close friends first. And I admit I'm kinda sensitive to negative feedback.

Comment by Ramiro P. (ramiro-p) on How much delay do you generally have between having a good new idea and sharing that idea publicly online? · 2020-03-17T22:00:45.903Z · LW · GW

Maybe a spreadsheet would be more informative. You could easily aggregate many answers.

Comment by Ramiro P. (ramiro-p) on Some quick notes on hand hygiene · 2020-02-11T18:26:30.712Z · LW · GW

If you grab your mobile with your dirty hands, then wash them, and then use your device again, you just recontaminated them; and if you never clean its surface (how do we do it effectively?), it'll accumulate pathogens. This seems to be a serious problem in hospitals.

(I'm not sure if I follow your reasoning; it apparently implies that, if you never shake hands with someone else, you never have to worry about washing them. Of course, it does reduce the potential for transmission.)

Comment by Ramiro P. (ramiro-p) on Some quick notes on hand hygiene · 2020-02-10T17:06:31.502Z · LW · GW

We should take into account the welfare of others, too. Besides protecting me from disease, washing my hands prevents me from transmitting it to someone else. It's pretty much analogous to vaccines.

Comment by Ramiro P. (ramiro-p) on Some quick notes on hand hygiene · 2020-02-10T17:00:36.387Z · LW · GW

I 've seen somethings concerning how dirty cellphones are - and how they can worsen interpersonal disease transmission. I wonder if there's any advice on how to keep it clean (and how useful would it be).

Comment by Ramiro P. (ramiro-p) on Why are people so bad at dating? · 2019-12-05T05:24:23.493Z · LW · GW

I do think Lanrian nailed it: there's no process ensuring fitness selection in dating. On the other hand, we are wasting an opportunity to go meta here: if everyone were capable of mimicking the features of a picture that make it successful, then those features should lose their importance, since they are not reliably signaling that someone is a good mate. If I'm not bright enough to see through a carefully planned image of a smile and a discrete cleavage, I am probably not bright enough to get a similarly attractive picture for myself. Plus, usual cognitive bias: I often mistake how much I like my pics (for the emotions they evoke, for example) with my assessment of their quality - and so irrationally rate them above average. But yeah, now i can use Photofeeler

Comment by Ramiro P. (ramiro-p) on Debate on Instrumental Convergence between LeCun, Russell, Bengio, Zador, and More · 2019-10-22T17:14:03.365Z · LW · GW

I find LeCun's insistence on the analogy with legal systems particularly interesting, because they remind me more Russell's proposal of "uncertain objectives" than the "maximize objective function" paradigm. At least in liberal societies, we don't have a definite set of principles and values that people would agree to follow - instead, we aim at principles that guarantee an environment where any reasonable person can reasonably optimize for something like their own comprehensive doctrine.

However, the remarkable disanalogy is that, even if social practices change and clever agents adapt faster than law can evolve (as Goodhart remarks), the difference is not so great as with the technological pace.

Comment by Ramiro P. (ramiro-p) on Honoring Petrov Day on LessWrong, in 2019 · 2019-09-28T12:29:10.509Z · LW · GW

So far, LW is still online. It means:

a) either nobody used their launch codes, and you can trust 125 nice & smart individuals not to take unilateralist action - so we can avoid armageddon if we just have coordinated communities with the right people;

b) nobody used their launch codes, because these 125 are very like-minded people (selection bias), there's no immediate incentive to blow it up (except for some offers about counterfactual donations), but some incentive to avoid it (honor!... hope? Prove EDT, UDT...?). It doesn't model the problem of MAD, and surely it doesn't model Petrov's dilemma - he went against express orders to minimize the chance of nuclear war, so risking his career (and possibly his life).

c) or this a hoax. That's what I would do; I wouldn't risk a day of LW just to prove our honor (sorry, I grew up in a tough neighborhood and have problems with trusting others).

My point is: I think (b) and (c) are way more likely than (a), so I'd use the launch codes, and take the risk of ostracism, if I had them. I think it would yield higher expected-utility; as I said, I wouldn't risk a day of LW to prove our honor, but I'd do it to prove you shouldn't play Petrov lightly.

Please, correct me if I'm wrong.

P.S.: (d) this allows anyone to claim having launch codes and mugger others into counterfactual donations - which is brilliant.

Comment by Ramiro P. (ramiro-p) on Honoring Petrov Day on LessWrong, in 2019 · 2019-09-28T11:23:55.680Z · LW · GW

I thought he was being ambiguous on purpose, so as to maximize donations.

Comment by Ramiro P. (ramiro-p) on Deducing Impact · 2019-09-26T00:34:16.048Z · LW · GW

I am wondering about the link between the notion of distance (in the first post), extremes in a utility scale, and big deal. That's me in 15'

Comment by Ramiro P. (ramiro-p) on Peter Thiel/Eric Weinstein Transcript on Growth, Violence, and Stories · 2019-09-04T20:55:31.108Z · LW · GW

My opinion ("epistemic status"): dunno.

I remember an issue in The Economist in 2013 about it. There's some argument among economists on the absence of productivity improvements, despite the buzz over AI and ICT; Erin Brynjolfsson argues that it takes some time for global pervasive technologies to have an impact (e.g.: electricity). However, the main point of Thiel & Weinstein is that we haven't found new breakthroughs that are easy to profit from.

But it reminds me Cixin Liu's Dark Forest context, where:

humankind stalled because Physics breakthroughs were prevented by the Sophon Barrier - even so, they built a utopian society thanks to cheap energy from fusion power.


Comment by Ramiro P. (ramiro-p) on How Much is Your Time Worth? · 2019-09-02T14:50:22.324Z · LW · GW

Your argumetn is sound. For me, it's curious that development economists almost never mention the temperature x productivity relation - except for J. Sachs (who mixes it with other geographical factors) and Nordhaus (who got a Nobel Prize for reasoning about it).

Comment by Ramiro P. (ramiro-p) on What questions about the future would influence people’s actions today if they were informed by a prediction market? · 2019-07-21T17:41:43.389Z · LW · GW

Ok, but the point is: how do you aggregate this in a prediction market? You have no incentive to bet on Earth's doom

Comment by Ramiro P. (ramiro-p) on Self-confirming predictions can be arbitrarily bad · 2019-06-07T01:26:20.536Z · LW · GW

There's a cool name for this donor's action: blindspotting (yeah, it's written like this) - after a Roy Sorensen book from 1988.