retsibsi

You're taking that phrase out of context. If you look at the parent and grandparent comments of the one you originally replied to, it's pretty clear that "the answer is profit" was pointing to the insinuation that Anthropic is deliberately causing Claude Code to waste lots of tokens for no reason other than short-term financial gain, and that's what solidasparagus was disagreeing with.

Are you talking about fixing conducted by the exchange / bookmaker, or just gambling-related fixing in general? The latter is of course happening, and I'm sure you're right that there will eventually be more public scandals. (And I agree with you that sports betting is very socially harmful.) But I thought the conversation was about fixing conducted by legal betting exchanges (sorry, 'prediction markets'), and I don't think your claims hold in that context.

> literature, you can trivially read all the classics by the time you are 30

I don't think this is true, at least not for an ordinary person, unless you're dedicating your life to it (hardly trivial) or ploughing your way through for the sake of completion rather than absorbing them intellectually and emotionally. Writers like Joyce can be difficult to even parse if you're trying to read their work quickly, and to truly understand it you'll probably need to engage with secondary sources. And then there's the likes of Proust, which aren't exactly difficult to read but are extremely long (and probably won't do much for you unless you take your time with them). There's no objectively valid set of 'classics' but we're talking about certainly hundreds and arguably thousands of books. A book a week every single week from the age of 15-30 would do it if we're taking about a medium-sized set of classics...

> Calling that an opt-out is so rich, like calling suicide an opt-out method. 24 hour waiting period to do something on a device I own.

What?

You only have to change that setting once, and you can do it in advance (before the restriction takes effect). So you don't even have to wait 24 hours at all.

I can see this change having negative effects on the android ecosystem, but the direct impact on your freedom to install apps on your device is basically zero. If that's not the case, please tell me what I've missed.

Metric rules and imperial drools, I agree, but I reckon if we put our minds to it we can multiply by 0.9 and 1.6 (or 0.91 and 1.61 if you are genius). Keeps the mind sharp in this age of outsourced thinking!

You say this like it's obvious, but I don't see the logic. When free users waste tokens, this costs Anthropic money. When subscribers waste tokens, this probably also tends to cost Anthropic money; the only way it benefits them is if it causes the user to buy a more expensive plan (or open a second paid account). If you or your company are paying per token, then sure... but is the amount of money they're making via this kind of wastage really enough to outweigh the risk of driving users away?

How many people are spending tokens on this in a way that is valuable to Anthropic? Free user tokens are a cost; subscriber tokens are either a cost or neutral, unless they cause the user to hit their limit AND this causes the user to buy a more expensive plan. And each customer they annoy might go to a competitor. So I think they almost certainly have an incentive to fix this, unless almost nobody cares about it.

I may have it wrong, but I think 'architecture' can include relatively low-level details: you could even design a program down to the level of what each function does, and still only be doing 'architecture', not coding, so long as you're not writing the code or specifying exactly how it should work.

You may be right that they are in a bind wrt public opinion. But that is not a defence against any specific criticism. What is your opinion on the conflict of interest mentioned above?

The point is that we don't know how to ensure they are aligned with anyone: not the company, the user, or the rest of humanity.

If someone claims that a model is 'aligned', it makes sense to insist that they specify with whom it is aligned. If someone claims that 'alignment is far from solved', they're probably pointing out that nobody understands the models well enough to be confident they are robustly aligned with any chosen set of values (including 'do whatever lukewarm707 wants').

Hah, I just watched it on your recommendation. I think they were going for the vibe of a cynical sitcom... but they forgot to put any decent jokes in, and they didn't do what successful sitcoms do and paper over the nastiness with charisma and some moments of warmth. So it's just 3 minutes or so of stress-inducing bleakness.

I definitely don't think it's unique to academia. I think some systems are more tethered to reality than others, though -- some are more like a ship, where even though the various crew members' interests are far from perfectly aligned, they all have a strong shared interest in not drowing, and most of them have a shared interest in getting to the intended destination efficiently. So although the system might become brutal and unequal, it's unlikely to shift toward a set of processes and norms that don't actually work at all.

My impression is that large portions of academia (and not just the most obvious candidates) are too close to the other extreme, with nothing to guarantee that indefinite cargo culting is not a stable equilibrium. And even when people notice this and get mad about it, they're more likely to flip the political sign of the problem than to actually fix it. But I hope this is overly cynical.

I think there's an equally large meta-incentive problem: the people who have power within the system are generally incentivised not to rock the boat, and the people who would like to rock the boat generally don't have power. This bind could be broken by external pressure (from government, itself incentivised by popular pressure) but it's very hard for most outsiders to understand and evaluate proposed changes, so even if the will was there, I don't know how much would be accomplished.

As an interested outsider myself, it's pretty depressing seeing how high the risk:reward ratio seems to be for people within academia who are driven to clean it up and call out bad actors, and how high the reward:risk ratio is for those who are willing to cut corners (or just plain cheat, if they do it in a minimally competent way).

I don't know HN's rationale, but: if you try to ban self-promotion, then in practice you're only banning honest self-promotion. And if you allow it, but try to ban accounts that only exist for that purpose, then in practice you're encouraging them to spam mediocre comments in other threads to dilute the signal. [edit: I see that the guidelines do say not to use HN primarily for promotion. I'm willing to bet this significantly increases the amount of insincere engagement, but obviously the admins/moderators here are more aware of the tradeoffs than I am, and they have decided they are worth accepting.]

IMO the important thing to try to crack down on is vote manipulation; if you have that under control, then you can probably happily allow self-promotion and trust that only the interesting stuff will rise to the top.

As I read it, that was sort of their point: if the theorists were blindly focused on Big O complexity, they could 'cheat' by invoking the universal search algorithm as a good-enough solution to a huge class of problems; but they don't, because they're not mindlessly playing a game where Big O complexity is all that matters.

> If anyone would like to present a secret third thing, please do so.

(c) The argument, as presented, can be read more than one way, and by 'steelmanning' it you are choosing to read it charitably and respond to its strongest version.

I think that's basically what the other guy meant: we can read "the world should accept some bad things" to mean something like "fuck you, we'll do what we like and the world will suffer the consequences" (less charitable, arguably a strawman) or something like "there will not be literally zero costs, but we think the shared benefits will be much greater" (more charitable, arguably a steelman), and he was choosing the latter.

I may be misunderstanding you, but I think our point of confusion/disagreement is that you're still reading them as talking about how Lean proofs are created, whereas, given the context, I'm confident they were talking about how Lean proofs are reviewed (after being created by a human or LLM).

I hope there's no hard feelings, by the way -- I wasn't meaning to be nasty to you in that comment. I guess I thought your reading was a bit uncharitable, but I understand that misunderstandings happen (very much including on my end) and I didn't mean to make a big thing of it, so I'm sorry if I came across rudely.

The person you responded to wasn't saying that Lean magically generates proofs, and all you have to do to create one is feed the problem into Lean. They were saying that, _given a Lean proof_, you can trust Lean to have verified that the steps in the proof are valid, and all you need to do manually is confirm that the theorem was correctly encoded. (IDK if that's true or misses important nuances, but it's not the obviously silly claim you read it as.)

Are you confident that Trump's FBI and DOJ (and Trump-aligned state police/prosecutors; if you are a Trump supporter, please substitute in the names of your most distrusted democrats) would vigorously pursue all credible allegations of vote-buying and coercion in his favour? And even if you are, and they did, do you think more than ~half the country would believe it?

Claude often starts a sentence with 'But'. (There are several examples in my recent conversations, which I checked in case I was misremembering.) 'And' seems to be much less common, but it would use that too if instructed to do so as a humanising trick.

> The point I'm trying to make is that I agree with the parent comment to mine

But you did that by making a big, surprising claim for which you haven't provided any evidence. You might not intend for that to be the main thing people focus on, but they are right to be both interested and skeptical! If true, it's a big deal.

The closest thing I could find is this (https://www.thenation.com/article/archive/voters-predominant...), which was attributed to 'human error' rather than fraud, and was resolved by allowing the affected voters to cast provisional ballots. 'Human error' is a vague explanation, and for all I know it was a coverup -- but the onus is on you to provide some positive reason to believe that. Surely there were plenty of people motivated to investigate this, so it would be surprising if a bullshit explanation was just quietly accepted and no further public attention paid.

I think this is a tangent. The obvious problem with signing someone else's name to work they did not create and do not endorse is that, if you publish the image, you will be falsely attributing authorship to them without their consent. The intent (if any) behind the addition of the signature is a separate issue.

But you're not just saying they are insufficiently good at reasoning, you're saying they're (probably) not "doing actual reasoning". So we need to know how you are defining "actual reasoning".

I don't think the bar for an actual reasoner can possibly be 'always appropriately applies syllogism to arrive at correct conclusions', because in that case nobody in the world is an actual reasoner. And if your bar were 'sometimes appropriately applies syllogism to arrive at correct conclusions', it's hard to understand why the current generation of AIs doesn't meet it; they are clearly capable of doing so, at least to all outward appearances. (Maybe you think their apparently successful demonstrations of reasoning are illusions, but again, you would need to define what counts as "actual reasoning" vs. a superficially convincing simulation of it.)

> Introspection is an internal self-driven process.

I don't think introspection is necessarily self-driven. If a psychologist tells me to examine my thoughts, and so I do, am I not introspecting?

No; he led multiple times but was in 7th when his engine went.

> It was an ex posto facto thing to claim saying it was a lab leak was racist. I’m sure someone said that, however, no one serious was nor was it a common position.

I think you are misremembering.

> The lab-leak theory had been floated by President Trump and others in his circle in April last year. It was widely dismissed at the time as an unfounded conspiracy theory with racist undertones.

https://theconversation.com/covid-19-why-lab-leak-theory-is-... (2021)

> New York Times reporter Apoorva Mandavilli deleted a tweet Wednesday night that said the COVID-19 "lab leak theory" had "racist roots."

https://www.yahoo.com/entertainment/ny-times-covid-reporter-... (2021)

> The lab-leak theory came to legitimacy by a circuitous path. It was first auditioned by Donald Trump and Mike Pompeo shortly after lockdown started, but journalists were quick to distance themselves from its overtones of crude Trumpian racism. Nevertheless the theory gained new life last spring, when some of those same journalists — abetted by a chorus of concern trolls — began to question if they had been blinded by liberal partisanship.

https://www.nplusonemag.com/issue-42/politics/lab-leak-theor... (2022)

Some of this was a case of people failing to recognise that Reversed Stupidity Is Not Intelligence, reasoning (implicitly) like this:

Trump lies all the time + Trump said X => not X

instead of like this:

Trump says whatever suits him, regardless of the truth + Trump said X => no significant change to our credence in X

> Calling it the “Chinavirus” was obviously racist and weird.

That was a dumb Trumpism for sure, but separately from that, here in Australia, before the global spread of covid became too obvious to ignore, there was a sense among right-thinking people that even being concerned about it, let alone wanting to take any kind of action, was a manifestation of anti-Chinese racism / very literal xenophobia.

The point would be that it's open source. And "chances are you're allergic" translates to "your use case / expectations are unusual". But in my opinion the analogy doesn't hold well at all, because destroying a persistent undo file without warning is pretty obviously incorrect and unexpected behavior.

> If my neighbor mows my lawn for free, I'm not going to complain about his workmanship. If I want the job done a particular way, the solution is to do it myself, or pay someone to do it the way I like.

I think this is a decent analogy, but it works better the other way. If my neighbour offers to mow my lawn, I accept, and then he destroys the flower bed adjacent to the lawn, I will be upset, and I will have every right to complain about what he did. If he reacts by blithely dismissing my concern, then I certainly have the right (and arguably the obligation) to warn others that they should think twice about accepting his offers of gardening assistance.

Some expectations reasonably go without saying; "don't destroy my flower bed when mowing my lawn" is one, and IMO "don't destroy my data without a clear warning and a chance to back out" is another, though of course we might disagree about exactly where this does and doesn't apply.

They get to do more or less what they want with the software, and others get to say more or less what they want about that. What's wrong with deciding they've crossed a line, being bothered by that, and warning others? Nobody's trying to get the law involved, or maliciously retaliate against them, or anything like that.

> If the model is nothing more than the sum of its training data and regime, then the company is responsible for its behaviour.

What stops the company from being responsible regardless? They created this entity, it's running on servers they own or rent, and (in these cases) it's acting on their instructions.

If it's also conscious, then IMO that greatly broadens their moral responsibility, because now model welfare matters. But we're talking about their responsibility for the model's actions, and I don't see how this could be weakened by model consciousness, given all of the above. As for their legal responsibility, the models don't have legal personhood, so who else but the company could be responsible?

It gets more complicated when the person who sets the model in motion (i.e. prompts it) is a third party, but in cases of internal models committing cybercrime during testing, surely the locus of responsibility is obvious.