wulfstan

Technically correct (the best kind of correct), but if I set a thousand users on to a website to each download a single page and then feed the information they retrieve from that one page into my AI model, then are those thousand users not performing the same function as a crawler, even though they are (technically) not one?

If it looks like a duck, quacks like a duck and surfs a website like a duck, then perhaps we should just consider it a duck...

Edit: I should also add that it does matter what you do with it afterwards, because it's not content that belongs to you, it belongs to someone else. The law in most jurisdictions quite rightly restricts what you can do with content you've come across. For personal, relatively ephemeral use, or fair quoting for news etc. - all good. For feeding to your AI - not all good.

He's that widely respected that amongst those who repair bikes (I maintain a fleet of ~10 for my immediate family) he is simply known as "Saint Sheldon".

Yeah I'm not so sure about that.

If Perplexity are visiting that page on your behalf to give you some information and aren't doing anything else with it, and just throw away that data afterwards, then you may have a point. As a site owner, I feel it's still my decision what I do and don't let you do, because you're visiting a page that I own and serve.

But if, as I suspect, Perplexity are visiting that page and then using information from that webpage in order to train their model then sorry mate, you're a crawler, you're just using a user as a proxy for your crawling activity.

This is not one person complaining about another’s behaviour - these are two people who are alleging the same thing about the same person. Either a) they are colluding or b) this is a pattern of behaviour by the accused person that needed to be addressed.

Whether the way it was addressed was appropriate or not I don’t know. The person may have been warned about it and what we are reading about may have been the end of a lengthy process. Maybe, as the accused person alleges, this was an enormous unexpected surprise and they had no warning of it.

You’re entitled to your opinion about what has taken place in this situation.

> OK, but this is pure speculation. I prefer not to bear such in mind.

Quote from the second accuser:

Because of similarities in our stories as well as the stories Jon Pretty told us himself, we have reasons to believe that other people in the community have had experiences with him that are similar to ours.

I think either you aren’t reading the accusers’ statements closely enough or you are choosing to discount them - but if they suggest that this is more widespread than just the two of them, I think this moves beyond “pure speculation” and into “quite likely”.

I think it’s selection bias - a y-combinator forum is going to have a larger percentage of people who are techno-utopianists than general society, and there will be many seeking financial success by connecting with a trend at the right moment. It seems obvious to me that LLMs are interesting but not revolutionary, and equally obvious that they aren’t heading for any kind of “general intelligence”. They’re good at pretending, and only good at that to the extent that they can mine what has already been expressed.

I suppose some are genuine materialists who think that ultimately that is all we are as humans, just a reconstitution of what has come before. I think we’re much more complicated than that.

LLMs are like the myth of Narcissus and hypnotically reflect our own humanity back at us.

I'm afraid I don't accept that you can split this into "experiencing something traumatic" and "making the accusation that you have experienced something traumatic".

The claim that "almost the entire community immediately sided with them" is accepting the accused's account of what happened in favour of the accusers. At least one of the victims started to raise concerns in the community several years beforehand and their concerns were not taken seriously:

"I have reported all of my experience to the ScalaCenter in 2019. I was hoping to see concrete actions, such as building a reporting mechanism, to protect minorities in the community. Unfortunately, I am not aware of such actions taken."

I'd also be very, very deeply skeptical that two public claims were the only claims made. We should bear that in mind. If the accusations are true, the public ones are usually the tip of the metaphorical iceberg.

> But this is not the case at all, unless you intended "these kinds of accusations" to mean both making formal charges and writing accusatory blog posts -- but the whole reason for this article is to point out the massive amount of damage that the latter can do at almost no cost to the accuser. Absent further evidence, it's clear that in this particular case, the two accusers' lives were not at all "torn apart" by making these life-destroying accusations -- do you agree?

Absolutely not! Assume the alleged victims are telling the truth, and read their statements again, carefully. Do they sound to you like people whose lives weren't torn apart by the experience? They needed counselling, therapy, time off work. These sound to me like traumatised people. You can argue that what they had to deal with wasn't "as bad" as what the accused had to deal with, but I don't accept that women make public accusations of sexual exploitation casually without any personal consequences, and certainly not in this case.

The "1 in 100" statistic is to remind people of a few things: firstly, knowing that you will have to expose your sex life to the police and there is only a very small probability that anything will actually be done about it, some women are still brave enough to try, and secondly, that underneath these 1 in 100 accusations are many others who just cannot bring themselves to the point of talking to the police about what they have experienced.

I think we should give women who make these accusations the benefit of the doubt while establishing the facts, acknowledging that coming forward to raise your voice about these things is extremely difficult. If men can by and large rape women - commit a crime against them - with relatively little risk of successful prosecution, then I think it's pretty obvious that non-criminal sexual exploitation is even less likely to have any consequences for the perpetrator.

If both people are telling the truth, then it sounds like you're saying that although very sad, a community "gatekeeper" sexually exploiting a vulnerable newcomer is just part of life and we should move past it.

I'm not sure I agree with this, and I think we can and should do better.

No - and in fact in my view this is the core problem with these kinds of situations - there isn't a long established process validating a set of accusations, that if true, fall short of criminality but should result in your exclusion from a community.

Individual communities have to establish ground rules for these sorts of things to protect the vulnerable.

> And now history is replete with examples of woman destroying the lives of men with no process or consequence.

I do not accept that this happens with nearly the regularity that people, usually men, claim it does. To make these kinds of accusations as a woman tears your life apart in unimaginable ways.

By way of example, 1 in 100 rape accusations MADE TO THE POLICE in the UK leads to a charge being made against the accused. That is what we as a society are up against, and why we have to take creepy, exploitative behaviour that falls short of criminality so seriously.

I do not accept that this is "naked bias".

If the accusations are true, then this is yet another example of a pattern of behaviour played out so regularly, across cultures, centuries and communities, that it is boringly predictable: "Senior community member, almost always a man, sexually exploits vulnerable women seeking acceptance into that community."

When a possible situation arises you should investigate it and, if there is reasonable evidence that it is true, do what you can to stamp it out and ensure it stops happening.

This is exactly right. Criminality is a very high bar! There are many behaviours that fall well short of criminality that we shouldn't accept in communities.

Criminality is a very, very high bar for removing people from a community for misbehaviour, and much sexual misconduct isn't criminal. I don't think "leaving things to the police" is good advice in situations where a vulnerable minority group needs to be protected from predatory behaviour.

In this situation you have an accusation of misconduct made by 1) young 2) women 3) new to a community 4) who don't speak English well. These are all big red vulnerability flags.

I would ensure that every accusation of this nature is treated with respect and investigated by a trusted authority figure in a given community.

Yes; that is correct. They are admitting they made an accusation without evidence or an investigation to support the claim. This is a civil matter and says absolutely nothing about the truth or otherwise of the claims made.

I have no idea who is telling the truth in this situation, and unless you are the person who has been accused or those who are the alleged victims, neither do you. For situations like this where the allegations fall short of criminal misconduct, a thorough process run by someone independent of the situation needs to a) to evaluate the claims made b) determine whether they are justified c) issue a clear and open report on what took place for the benefit of the community involved. As far as I can tell no investigation has been carried out to verify or falsify claims made by the individuals concerned.

But - it is worth stating very clearly that history is replete with examples of men who have used their senior position in communities to take advantage of women, and if what these women say is true, it would be utterly unsurprising to me. The High Court judgement in this situation is a civil matter; nobody has been "cleared" of anything.

In the absence of an investigation, you can read the original statements made by the women who made the accusations of wrongdoing [here](https://medium.com/@yifanxing/my-experience-with-sexual-hara...) and [here](https://killnicole.github.io/statement/), and you can form your own opinion about who is telling the truth based on what little there is to go on.

EDIT: s/judgement/opinion/

If you're tired of this BS can I encourage you to make:

https://noai.duckduckgo.com/

Your default search engine. AI overviews begone.

If that ends up being the case then we could all genuinely agree that good has eventually emerged from the compute/inference infrastructure that LLMs paid for. I hope that comes to pass.

In July 2023, I wrote this to a friend:

"...being entirely blunt, I am an AI skeptic. I think AI and LLM are somewhat interesting but a bit like self-driving cars 5 years ago - at the peak of a VC-driven hype cycle and heading for a spectacular deflation.

My main interest in technology is making innovation useful to people and as it stands I just can't conceive of a use of this which is beneficial beyond a marginal improvement in content consumption. What it does best is produce plausible content, but everything it produces needs careful checking for errors, mistakes and 'hallucinations' by someone with some level of expertise in a subject. If a factory produced widgets with the same defect rate as ChatGPT has when producing content, it would be closed down tomorrow. We already have a problem with large volumes of bad (and deceptive!) content on the internet, and something that automatically produces more of it sounds like a waking nightmare.

Add to that the (presumed, but reasonably certain) fact that common training datasets being used contain vast quantities of content lifted from original authors without permission, and we have systems producing well-crafted lies derived from the sweat of countless creators without recompense or attribution. Yuck!"

I'll be interested to see how long it takes for this "spectacular deflation" to come to pass, but having lived through 3 or so major technology bubbles in my working life, my antennae tell me that it's not far off now...

I keep saying to people - "if you have a good idea that can make use of large amounts of really really cheap GPUs to do something genuinely useful - get ready for a massive glut of spare capacity". I still haven't thought of anything, unfortunately...

> Except not every society is such a compromise. Some are fully under authoritarian control, and serve as a warning for others who are tempted by authoritarian ideas.

Every society is on a continuum, and so represents some compromise between freedom for the citizen and power for the authorities. No society is perfectly free and no society is entirely authoritarian.

> Breaking encryption need not inherently be part of that compromise. And until someone can explain how breaking encryption will actually stop the kind of bad actors used to justify such a direction (vs. driving them deeper underground, i.e. even if you outlaw encryption, it’s not as if law breakers will obey such a law), I see no merit in entertaining such a compromise. The crimes being committed are already illegal.

Being able to legally access a private citizen's encrypted data in specific situations would help to (at least more rapidly) prosecute certain crimes more successfully. This is, I think, inarguably true. You can decide for yourself if that is worth a compromise. I'm somewhat on the fence.

> I don’t think most people in our community see it as inherently/perfectly good, but as extremely important and necessary. This is a critical distinction. As with everything, there are harms that come with the good, and such is the nature of all things. The question becomes: are the harms allowed worse than the good that is preserved? And would the new harms of disallowing the status quo be potentially worse than the harms supposedly prevented?

I think it's convenient and useful, but I hardly think it's necessary. Society managed to function just fine (although less conveniently) when strong encryption wasn't available for communications. Banking still happened, money still changed hands.

> I agree that we need to think critically about this. But clearly we disagree about what one should conclude from such a critical analysis. I’d argue that taking the position that the government needs more power - especially at this moment in history - is the result of not thinking critically enough.

It depends on how you define power. As society changes and new technologies emerge, maintaining existing government authority in new areas - and working out ways to ensure that authority is maintained - isn't really giving governments more power, but trying to ensure your society remains in the agreed location on the freedom/authority continuum.

If you see this as expanding powers, I can see how you would consider that a problem. But I think this is more about ensuring existing power is maintained correctly over a new area where crime is being committed.

It requires decryption by means beyond the sole supply of a user-owned key. That doesn't require a manufacturer to be capable of decrypting it.

I suspect that there are many ways that can be achieved, all technical ;-)

> Do you think that this can be done without introducing massive security weaknesses into systems that cannot have them?

Yes, I do - or rather, that is the point of the discussion. We currently allow central authorities to indicate our permission to do or be something in the root certificate system. Why can't something similar be designed to allow controlled decryption?

> Also, there is a question if you believe the authorities that without decrypting data, they can't investigate crimes.

Clearly there are circumstances in which being able to decrypt the data of a criminal would assist in prosecuting crime. See EncroChat for an example of how this has worked.

> Imagine an analogical assertion that without torturing suspects, law enforcement is stymied. Someone might assert that, but we still say no, for all sorts of fundamental reasons. Same with American Miranda rights and others.

Yes. Clearly there are reasonable limits that need to be applied before we can allow controlled decryption of data. I am not arguing for issuance of a master key. See my original post.

> Myself, I don't believe in that assertion at all. Most crimes leave a massive real world trace that cannot be encrypted. The ones that don't, maybe should not be crimes in the first place.

Some do, and some don't. Things like e.g. cryptocurrency heists have profound effects, and are propping up North Korea. Those are definitely crimes...

I get the hostility towards it - as I've said elsewhere, it's practically an article of faith in our community that strong encryption == unalloyed good. And clearly it needs a lot of thinking to address potential abuse. But we've done it for other things.

> Not to be too pedantic, but I think the distinction between privacy and encryption is incredibly important: almost everyone agrees that privacy is a gradient. The disagreement is whether or not encryption can be a gradient. Most people do not think it can reasonably be without undermining ~everything relying on it.

That is a fair criticism. I would answer that by saying that encryption is just a technology, and you can employ it in very flexible ways (including e.g. n-of-m style keys) which if thought through well and legislated carefully could give the authorities more reasonable access to data when it is legally warranted.

Again, as a thought experiment, what legal protections can we put in place - an encryption ombudsman or independent authority - that would allow an arms length, controlled and expiring mechanism that allows limited access to a user's data? What would we as a society be happy to accept? I don't think the demand is an unreasonable one, but I'm trying to figure out what a reasonable collection of mechanisms looks like.

Well that's your view - but these demands aren't going to go away, and what I think is sensible is for us a technical community to consider reasonable alternatives. Every society is a compromise between anarchic freedom and authoritarian tyranny, and this is another discussion about how a (relatively) new set of technologies can fit into that compromise in a way that is acceptable and reasonable.

I acknowledge the problems you raise, but it does seem to me that we have a good set of systems in place in the form of PKI that has a remarkable amount of flexibility.

It's frankly a bit of an article of faith in our community that encryption == unalloyed good and I think we'd be right to think more critically about that position.

Apple's iOS firmware encryption key hasn't leaked, as far as I am aware. Do you know otherwise?

There are already very good solutions for ensuring that key leakages are very difficult to do and limited in effect.

Yes, several times.

But in the same way that as a society we allow physical privacy (and freedom!) to be removed under certain circumstances, we should consider allowing digital privacy to be removed in the same way. 1984 imagines a world where the authorities can enter your physical space at any time because they feel like it. But I don't lie awake worrying about that because I live in a society where I feel the social contract is largely upheld by the authorities.

> In those cases, you need a judge or someone to approve the seizure. With a backdoor that can be opened at any time, you should consider that nothing will be private because there is no one who is going to be monitoring it 24/7 to make sure that there are no abuses.

I'm not sure you've read what I wrote correctly. My hypothetical system would not allow the backdoor to be opened at any time, but it would require a certificate to be issued (derived from the manufacturer / application creator's root) that gives limited, expiring access on the production of court-authorised warrant, in exactly the same way a judge gives the police permission to enter your physical property.

I'm not sure I follow. Obviously the application itself needs to support the business logic described, in the same way as your web browser needs to notice that a certificate has expired and tell you there's a problem with a website you're visiting. What I'm exploring is why requiring certain applications to support the same sort of thing to decrypt user data in certain circumstances to support law enforcement is a problem.

Then setup the system to be more analogous. Make the publication of key issuance under this mechanism public after a period of time.

Again, I see us falling back into an "all or nothing" view of privacy and I just don't think those are the only options.