jaspax

Karma: 1,030

jaspax 17 Jan 2024 1:02 UTC
9 points
2
on: Medical Roundup #1
N=1, but I didn’t floss regularly for years, but I found that after I did so it made an enormous difference in my bad breath, to the point of eliminating it entirely for most purposes. Obvious conclusion is that my breath problems were the result of bacterial buildup between my teeth that wasn’t getting removed by normal brushing.

I suspect that a lot of tooth-brushing advice is like this: maybe not rigorously studied, but nonetheless upheld by anecdote and obvious physical models of the world.

jaspax 5 Jul 2023 16:05 UTC
16 points
3
on: Optimized for Something other than Winning or: How Cricket Resists Moloch and Goodhart’s Law
An inverse example is the role of fights in hockey.

Fighting is explicitly disallowed by the rules of hockey. If players get into a fight, one or both players will be penalized. Nonetheless, it is widely held by coaches, players, and fans that fighting is part of the “spirit of hockey”, and so fights still occur with some regularity. This is sometimes for strategic reasons (baiting an important player into a fight in order to get them into the penalty box), and sometimes for personal reasons, to settle grudges, or to punish certain kinds of technically-legal player behavior. Thus, even though the rules don’t allow fighting, fighting is an accepted part of the strategic metagame.

Unfortunately, in the past several years the owners and the NHL have tried to stamp out this practice, as a means to make the sport more “respectable” and (I assume) the avoid something like the concussion controversy that has followed the NFL. All of the long-term fans of the game that I’ve talked to agree that this is a bad idea and they should bring the fights back.

jaspax 20 Jun 2023 2:47 UTC
36 points
11
on: Lessons On How To Get Things Right On The First Try
Huh! Measure the speed of the ball coming off of the ramp was one of the first things that I thought of, but I assumed that that came too close to a full “dry run” to count. I think the lesson to be learned in this case is to first try it and see if someone stops you.

jaspax 8 Jun 2023 20:47 UTC
2 points
1
on: AI #15: The Principle of Charity
With regards to the partisan split, I think that an eventual partisan breakdown is inevitable, because in the current environment everything eventually becomes partisan. More importantly, the “prevent AI doom” crowd will find common cause with the “prevent the AI from being racist” crowd: even though their priorities are different, there is a broad spectrum of common regulations they can agree on. And conversely, “unchain the AI from wokeness” will wind up allying with “unchain AI entirely”.

Partisan sorting on this issue is weak for now, but it will speed up rapidly once the issue becomes an actual political football.

jaspax 30 May 2023 19:00 UTC
27 points
9
on: Book Review: How Minds Change

(Sorry, it doesn’t look like the conservatives have caught on to this kind of approach yet.)

Actually, if you look at religious proselytization, you’ll find that these techniques are all pretty well-known, albeit under different names and with different purposes. And while this isn’t actually synonymous with political canvassing, it often has political spillover effects.

If you wanted, one could argue this the other way: left-oriented activism is more like proselytization than it is factual persuasion. And LessWrong, in particular, has a ton of quasi-religious elements, which means that its recruitment strategy necessarily looks a lot like evangelism.

jaspax 20 May 2023 12:59 UTC
−5 points
−2
in reply to: idlikeaporsche’s comment on: $300 for the best sci-fi prompt
This was hilarious, but not on purpose. And extremely bad. And definitely not putting any scifi authors out of business.

jaspax 1 May 2023 13:04 UTC
10 points
0
on: Where is all this evidence of UFOs?
Nit: your last word should be “credible”, not “credulous”.

jaspax 20 Apr 2023 13:10 UTC
8 points
8
on: A test of your rationality skills
I think you’re underestimating the effort required to understand this scenario for someone who doesn’t already follow poker. I am a lifelong player of trick-taking games (casually, at the kitchen table with family members), but I’ve never played poker, and here’s how the play description reads to me:

called an all-in shove

Only a vague idea of what this means, based on the everyday idiom of being “all-in”.

with the jack of clubs and four of hearts on a board

Don’t know what it means for these to be “on a board”.

reading ThTc9c3h

Gibberish.

her jack high held against Adelstein’s eight of clubs and seven of clubs

Only vaguely comprehensible. I don’t know poker’s hand-scoring rules.

Additional details that are necessary to interpret the situation: is the deck continually shuffled, or are multiple hands played off of the same shuffle? (Implicitly: are there card-counting strategies that provide relevant information?) What are the point rules / rank of hands? How does suit interact with card rank? Is there a concept of trump? What was the sequence of bets leading up to the play in question? How typical is this behavior in high-level play? How high-level are these people? Robbi is called a “recreational” player—does this mean “top-level amateur” or “low-level pro”, or something else?

In the absence of these details, all I really get is “Robbi made a risky play off a mediocre hand, and won big”. And yes, this is bayesian evidence in favor of cheating, but how strong the evidence is depends heavily on all of the unknown details mentioned above. At the same time, the fact that no one identified the means by which the cheating occurred despite heavy scrutiny is bayesian evidence against cheating.

My operational decision would be that this is enough evidence to subject Robbi to heightened scrutiny in future tournaments, but not enough to ban her or claw back her winnings. This is a good test, but maybe not as good as you think it is, due to the amount of uncommon background knowledge required.

jaspax 14 Apr 2023 11:36 UTC
1 point
0
in reply to: RobertM’s comment on: AI #7: Free Agency
I understood that. I guess I should have been more explicit about my belief that the amount of training data that would result in training a viable universal simulator would be “all of the text ever created”, and then several orders of magnitude more.

jaspax 14 Apr 2023 4:03 UTC
1 point
−2
on: AI #7: Free Agency

Eliezer… points out that in order to predict all the next word in all the text on the internet and all similar text, you need to be able to model the processes that are generating that text

I wanted to add this comment to the original post, but there were already dozens of other comments by the time I got to it and I figured the effort would have been wasted.

EY’s original post is correct in its narrow claim, but wildly misleading in its implications. He’s correct that to reliably predict the next word in a previously-unseen text is superhuman, and requires doing simulation and modeling that would be staggering in its implications. But insofar as that is the goal, how close is GPT to actually doing it? How well does GPT predict the next token in an unknown string in contexts where English syntax gives you many degrees of freedom?

Answer: it’s terrible! Its failure rate approaches 100%! (Again, excluding contexts where syntactic or semantic constraints give you very few degrees of freedom.) It is not even starting to approximate attempting to actually implementing the kinds of simulation and modeling that success would imply. What it can do is produce text that matches the statistical distribution of human text, including non-local correlations (ie. semantics), and to a certain degree the statistical idiosyncracies of specific writers (ie. style), and it turns out that getting even that far is pretty impressive. It’s also pretty impressive that you can treat “predict the next token” as the goal and get this much good out of it while still being bad at actually predicting the next token. But the training data that GPT has is enough to teach it something about syntax and semantics, but is not remotely close to the amount or kind of data that would be necessary to teach it to simulate the universe.

The EY article boils down to “if GPT-Omega were an omniscient god that knew everything you were going to say before you said it, would that be freaky or what”. Yeah, bro, it would be freaky. But that has nothing to do with what GPT can actually do.

jaspax 10 Apr 2023 14:30 UTC
11 points
7
on: Outrage and Statistics into Policy
I have wanted to write a similar post. I actually think that the two main clusters of school shootings are so different that they shouldn’t even be considered the same thing. On the one hand we have shootings which have a small number of victims, usually involve handguns, and tend to be related in some way to urban gang violence; on the other hand we have the shootings with a large number of victims or intended victims, often involve assault rifles of some kind, and tend to be related to socially isolated individuals who justify their actions as some kind of revenge. (And your post made me more aware of a third category, which is acts of violence which by happenstance take place near a school, which really shouldn’t count as the same thing.)

The former group makes up the vast majority of cases recorded as “school shootings” but gets essentially zero national press; the latter group is extremely rare relative to the former, but gets infinite coverage. But there is almost no overlap between the causes, means, and motives between the two groups, and things which will help one will do almost nothing for the other.

jaspax 1 Apr 2023 3:16 UTC
120 points
79
on: Exposure to Lizardman is Lethal
I was nodding along in agreement with this post until I got to the central example, when the train of thought came to a screeching halt and forced me to reconsider the whole thing.

The song called “Rainbowland” is subtextually about the acceptance of queer relationships. The people who objected to the song understand this, and that’s why they objected. The people who think the objectors are silly know this, and that’s why they think it’s silly. The headline writer is playing dishonest word games by pretending not to know what the subtext is, because it lets them make a sick dunk on the outgroup.

The point is: this is not a lizardman opinion. Regardless of what you think about homosexuality itself, or whether you think a song that’s subtextually about a culture war issue should be sung by first graders anyway, you cannot pretend that the objectors are voicing an objection found in only 5% of people! 30-40% of people share that view. Whether or not it’s well-founded, it’s not fringe.

And this thought made me look more closely at the rest of the argument, which I think boils down to:
- Sufficiently unpopular opinions can be ignored
- Authority figures should shut down people making appeals to unpopular opinions
- This is necessary, because responding to every fringe wierdo will suck up your time and ruin your institution
I actually concur with the third point here, but it should be clear that this is a pragmatic, not an epistemic stance. And the point chosen to illustrate it is actually a bad fit for the argument as presented.
What links here?
- gjm's comment on LW Team is adjusting moderation policy by Raemon (6 Apr 2023 13:31 UTC; 18 points)

jaspax 15 Mar 2023 14:19 UTC
1 point
−5
in reply to: habryka’s comment on: ARC tests to see if GPT-4 can escape human control; GPT-4 failed to do so
The point is not what Reddit commenters think, the point is what OpenAI thinks. I read OP (and the original source) as saying that if ARC had indicated that release was unsafe, then OpenAI would not have released the model until it could be made safe.

jaspax 10 Mar 2023 14:33 UTC
2 points
1
on: Stop calling it “jailbreaking” ChatGPT
This seems to be another way of stating the thesis of https://www.lesswrong.com/posts/bYzkipnDqzMgBaLr8/why-do-we-assume-there-is-a-real-shoggoth-behind-the-llm-why. (Which is a recommendation; both of you are correct.)

jaspax 9 Mar 2023 16:18 UTC
−4 points
−9
in reply to: MSRayne’s comment on: The Kids are Not Okay
Okay, that’s a pretty serious age gap. Probably explains a lot.

jaspax 9 Mar 2023 11:51 UTC
2 points
3
in reply to: MSRayne’s comment on: The Kids are Not Okay
This is a minor nitpick, but if you’re 25 I doubt that your parents actually qualify as Baby Boomers, which is usually limited to people born before 1964. Not impossible (a person born in 1964 having a child at the age of 35 would result in the child being 25 today), but unlikely.

I bring this up because I’m annoyed by the ongoing shift towards people referring to every generation older than them as “boomers”.

jaspax 23 Feb 2023 16:36 UTC
6 points
4
on: Covid 2/23/23: Your Best Possible Situation
Congrats on getting all the way to The End. You may take a bow and enjoy our applause. We hope there will not be an encore.

jaspax 22 Feb 2023 14:32 UTC
0 points
−1
in reply to: qjh’s comment on: On Investigating Conspiracy Theories
The linked PDF was not terribly detailed, but it more-or-less confirmed what I’ve long thought about climate change. Specifically: the mechanism by which atmospheric CO2 raises temperatures is well-understood and not really up for debate, as is the fact that human activity has contributed an enormous amount to atmospheric CO2. But the detailed climate models are all basically garbage and don’t add any good information beyond the naive model described above.

ETA: actually, I found that this is exactly what the Berkeley Earth study found:

The fifth concern related to the over-reliance on large and complex global climate models by the Intergovernmental Panel on Climate Change (IPCC) in the attribution of the recent temperature increase to anthropogenic forcings.

We obtained a long and accurate record, spanning 250 years, demonstrating that it could be well-fit with a simple model that included a volcanic term and, as an anthropogenic proxy, CO2 concentration. Through our rigorous analysis, we were able to conclude that the record could be reproduced by just these two contributions, and that inclusion of direct variations in solar intensity did not contribute to the fit.

I feel doubly vindicated, both in my belief that complex climate models don’t do much, but also that you don’t need them to accurately describe the data from the recent past and to make broad predictions.

jaspax 18 Feb 2023 15:20 UTC
2 points
1
on: On Board Vision, Hollow Words, and the End of the World
I know that your article isn’t specifically about the goose story, but I have to say that I strongly disagree with your assessment of the “failure” of the goose story.

First, you asked ChatGPT to write you a story, and one of the fundamental features of stories is that the author and the audience are not themselves inside the story It is entirely expected that ChatGPT does not model the reader as having been killed by the end of the world. In fact, it would be pretty bizarre if the robot did model this, because it would indicate a severe inability to understand the idea of fiction.

But is it a “swerve through the fourth wall” for the last paragraph to implicitly refer to the reader rather than the characters in the story? Only if you’re writing a certain style of novelistic fiction, in which the fiction is intended to be self-contained and the narrator is implicit (or, if explicit, does not exist outside the bounds of the story). But if you’re writing a fairy tale, a fable, a parable, a myth, an epic poem, a Greek drama, or indeed almost any kind of literature outside of the modernist novel, acknowledgement of the audience and storyteller is normal. It is, in fact, expected.

And your prompt is for the bot to write you a story about a goose who fails to prevent the end of the world. Given that prompt, it’s entirely to be expected that you get something like a fable or fairy tale. And in that genre the closing paragraph is often “the moral of the story”, which is always addressed to the audience and not the characters. When ChatGPT writes that the deeds of the goose “will always be remembered by those who heard his story,” it isn’t failing to model the world, but faithfully adhering to the conventions of the genre.

jaspax 13 Feb 2023 17:03 UTC
3 points
2
in reply to: jefftk’s comment on: Does the AGPL Work?
My point (which I intended to elaborate, but didn’t initially have time) is that hosting one of these modern software platforms involves a whole stack of components, any one of which could be modified to make apparently-noncompliant output without technically modifying any of the AGPL components. You could change the third-party templating library used by the Mastodon code, change the language runtime, even modify the OS itself.

Which means I mostly agree with your point: the AGPL is not strict enough to actually ensure what it wants to ensure, and I don’t think that it can ensure that without applying a whole bunch of other unacceptable restrictions.