I feel like a lot of the criticisms made here could be goal-factored out. I think Caplan would agree with them. But I think a lot of the things that are suggested would be missed out on (teaching conformity, peer groups, etc) could be replaced by significantly better alternatives than schooling.
This isn't exactly true, it's relatively common to see HN-user created games on the front-page. I think you're engaging in typical-mind; while this community is a space for nerds, not everyone here is interested in video games. The shared interest here is for "hacking", computer science, and computer software. I wouldn't even say the forum is necessarily business oriented, despite the association with Y-Combinator. It's simply that, because of the incentives, there is a lot of interesting software and research that comes from the private sector.
But we can watch it and learn and I don’t really see why not. I doubt we need to be so paranoid and see giving access to the internet to a LLM as so dangerous.
Less excitingly, an LLM with access to the web could do things with your online persona or IP that you’d find embarrassing or illegal. Maybe not when it’s slowed down and watched at all times, but will that always be the case once we start doing this?
Anyway the genies out of the bottle and “that’s an unsafe use of technology” is basically antithetical to the Silicon Valley ethos, so objecting at this point seems futile.
Theoretically an agent exposed to the internet could improve itself. But this one can not do that. There is no way (as far as we know) for anyone or anything with internet access to change the code running on GPT-4 short of finding out who works at OpenAI and blackmailing them. This would be easily detected.
You’re right that it could do something bad with your IP, but it’s not really correct to say that GPT-4 could improve itself if given internet access. It’s just not hooked up that way.
Relatedly, the original Colossal Cave Adventure, often known simply as "Adventure", which was a very early videogame and inspired so many others that followed (including the Zork franchise and Kentucky Route Zero) was loosely inspired by the creators' caving expeditions in Kentucky's Mammoth Cave (National Park) and some of the rooms and room descriptions in Colossal Cave Adventure even match up with real Mammoth Cave rooms. (Though my attempts to use the magic word XYZZY to shortcut to find Dragon hoarded gold at the real Mammoth Cave seem to always fail.)
This is ridiculous and has been a cascade of awful precedents for the Internet. This is legitimately an extremely bad sign of what's to come for the Internet's future.
Popular endorsement of harassment to the point of suicide seems like a worse sign, personally.
When it comes to threats to freedom of expression, I'm much more concerned about the religious right's abuse of government authority to censor education, ban books, and suppress the vote, than I am about private institutions taking steps to protect human life.
Private institutions have every right to stand behind whoever they choose. As a user, I have no intention of using Cloudflare because (even as a queer individual) I don't feel safe using a platform that would censor me if enough people were mad enough. I would much rather choose a platform that stands by their TOS as-written, instead of stepping in to arbitrate on a case-by-case basis.
Cloudflare has ever right to open Pandora's Box, but I want nothing to do with it. Much like Namecheap's fumble earlier this year, the way they handled this situation showed their true colors, and made it evident that I don't want to ever do business with them.
You'd think that, but historically this has only encouraged niche businesses to crop up catering to that crowd. Look at Epik (or Vultr) for example, businesses that exist solely to counter this kind of threat. Yes, they still reserve the right to remove abusive users, but that's defined by legal statutes and technical limitations rather than 'icky feelings'. Both services have a surprisingly solid track-record servicing the roughest of customers.
Failing all that, KiwiFarms doesn't need a business to stay afloat. The endgame for all of these so-called 'abusive platforms' is retreating to I2P/Tor, or another internet-adjacent network. To stop KiwiFarms from existing, you need to literally silence the people using it, not just shut down their clearnet website. Websites don't harass people, people do.
At the end of the day, businesses gotta eat. They're not a charity or a benevolent public force. If your user generated content impacts their bottom line, you're gonna get kicked to the curb.
>The endgame for all of these so-called 'abusive platforms' is retreating to I2P/Tor, or another internet-adjacent network
I'm pretty positive Cloudflare and IA do not care about a moral crusade to stop Kiwifarms, and would not care if they went to Tor. Both CDNs are primarily concerned about business risk with hosting content and calls to action that could be found illegal.
Epik hosted 8chan in the interim after Cloudflare dropped them. This cause Epik's hosting provider to drop them, and since Epik doesn't own their datacenters they had to abide by their hardware provider's decision: https://www.theverge.com/2019/8/5/20754943/8chan-epik-offlin...
I'll still eat some crow, though; I forgot the entire business was owned by Rob Monster (a veritable idiot by most definitions of the word), and I completely forgot that they even provided hosting in the first place.
> I'm pretty positive Cloudflare and IA do not care about a moral crusade to stop Kiwifarms, and would not care if they went to Tor.
I'm certain they don't. That's the problem, though; this moral panic response to KiwiFarms has achieved nothing. Cloudflare knows that this is a zero-sum game, but they bent anyways. As businesses, their choices make plenty of sense. I disagree with businesses all the time though (check the comment history), and frankly I think Cloudflare made the wrong decision here. In my opinion, their actions here will be more destructive to queer populations in the long-run.
It doesn't stop Nazis from existing though (or even knowing about, visiting and supporting the site). If we do the same thing with KiwiFarms, we just make it harder to monitor and easier for serial-abusers to collaborate. The majority of KiwiFarms users simply aren't solely enabled by the website existing, either.
I simply don't believe in deplatforming, and it disappoints me to see Cloudflare shrug and cave in.
> If we do the same thing with KiwiFarms, we just make it harder to monitor and easier for serial-abusers to collaborate.
Yes because the police were doing such a great job at monitoring and responding to threats from KF. Wake up, this is such a weak argument that has no basis in reality. We hear these same tired arguments over and over that are not backed up with facts (or, you know, arrests).
> The majority of KiwiFarms users simply aren't solely enabled by the website existing, either.
Yes they are.
> I simply don't believe in deplatforming, and it disappoints me to see Cloudflare shrug and cave in.
What part do you not "believe" in? Deplatforming absolutely works. These rats might scurry to another platform/provider but each time fewer and fewer jump through all the hoops. You are letting perfect be the enemy of good.
> It doesn't stop Nazis from existing though (or even knowing about, visiting and supporting the site).
This approach is making perfect the enemy of good. There is no perfect way to solve this problem.
> If we do the same thing with KiwiFarms, we just make it harder to monitor and easier for serial-abusers to collaborate.
It's still a website, it's still accessible. That part hasn't gone away. If the contention is that launching TOR somehow means it's "harder to monitor" then I have some news for you: Lots of monitoring of darkweb stuff goes on every day.
Deplatforming is analogous to quarantining a person with an infectious disease. It works. The FBI can still monitor whatever's left of KF on Tor or Discord.
If you think of beliefs as infectious diseases, you're not respecting the autonomy of other people.
If I need to be prevented from knowing what the monsters say in order to avoid turning into a monster myself, then I promise you, we've lost. The monsters are going to eat us all.
You've got plenty of autonomy left to visit them in whatever darkweb hellhole they decide to exist in and learn what the monsters are saying. We don't need to allow places in the public square for these people.
Yes, I can probably for myself still find out what they are saying. But this is about KF being removed from an archive site. Making it harder to prove what they are saying, and easier to lie about what they are saying. Who benefits from that?
>You'd think that, but historically this has only encouraged niche businesses to crop up catering to that crowd. Look at Epik (or Vultr) for example, businesses that exist solely to counter this kind of threat.
And Cloudflare had no problem hosting Kiwi Farms and Daily Stormer until they crossed a line. Cloudflare's history doesn't exactly paint it as bleeding-heart liberal who can't deal with 'icky feelings.' I'm sure Epik and Vultr have their lines as well, it just happens that none of their customers have crossed it yet.
>To stop KiwiFarms from existing, you need to literally silence the people using it, not just shut down their clearnet website.
Slowing them down is still a valid goal.
>Websites don't harass people, people do.
Guns don't kill people, people do. Except people with guns can kill a lot more people faster. That's why guns are a thing.
Whether or not you want to be a free speech absolutist, you have to concede that the platform and its reach matters. If it didn't, no one would be up in arms about deplatforming. Yes, it's literally and technically true that a website can't harass people, but having a platform meant to organize and facilitate harassment is a force multiplier for the people doing said embarrassment. Without the website, the people couldn't harass as well as they could with it.
And the size, reach and convenience of the network matters in that regard, just as the capacity, rate of fire and caliber of a gun matters, even if it is a person pulling the trigger.
> I'm sure Epik and Vultr have their lines as well, it just happens that none of their customers have crossed it yet.
Indeed they do, which makes me a happy customer. Knowing how inflammatory their other customers are, it brings me great comfort in knowing that their free speech is honored as much as mine. If either of them pulled a "Cloudflare moment" at the same scale, I'd probably start looking for other hosting providers.
> Slowing them down is still a valid goal.
...did we do that, though? The past 2 months have done nothing but put KiwiFarms in the spotlight. Instead of privately petitioning Cloudflare to change their policy, we drew battle lines and took to Twitter. All KiwiFarms ever wanted was attention, and we gave them more attention than they could have ever hoped for. Do people seriously think they're going to struggle to bounce back after an attack like this? Giving online organizations a platform has been a huge mistake in the past, like treating "Anonymous" as anything other than the default name for 4chan posters.
> you have to concede that the platform and its reach matters
Absolutely. That's why I'm afraid that attacking the platform now will cause it to become harder to attack. It's already increased it's reach, the recent media hubbub has ensured that everyone knows about KiwiFarms. I guess the Streisand effect is lost on modern internet users...
> Without the website, the people couldn't harass as well as they could with it.
Right. Now imagine how much worse things would get if there wasn't a website, but a Tor hidden service. Or a closed Matrix homeserver. Or an IPFS bulletin board. The sky is the limit, and I'd go as far as to argue that they were the least harmless on the surface web. Only time will tell, though.
> And the size, reach and convenience of the network matters in that regard, just as the capacity, rate of fire and caliber of a gun matters
Well... no. This is something that has been proven time and time again in America; banning certain types of guns doesn't work. Banning an AK doesn't stop someone from chopping their Glock 17 and clearing a room at half the price. Gun legislation doesn't correlate with a reduction in firearm violence. The capacity, rate of fire and caliber never mattered, just the fact that the gun existed in the first place. If we're not going to ban guns outright, what's the point in picking-and-choosing which ones are-and-aren't perceived as harmful?
Obviously it's a reductive argument, but the same thing goes for free speech. By choosing to draw the line somewhere, we're giving other people the go-ahead to draw different lines. We're giving world governments the tools they need to oppress LGBT users. We're drawing the blueprints for a new era of information suppression, and nobody seems to care since both sides have started beating the "muh terrorism" and "think of the children!" drums, respectively. And when has that ended well for internet freedom in the past?
i had ZERO clue what kiwifarms was before all this.. now that I know how much they try to keep me from seeing it, I have made it a point to look at for the forseeable future.
I do not subscribe to bullying or doxxing or anything like that, but now kiwifarms will get a fair chance at showing me what they are about, they had zero chance before, as I did not know they exist.
A website was bullied off the internet because it showcased the horrible activities of horrible people. I'm far more concerned that a handful of Twitter users can memoryhole an entire community.
You say that as if the people running KF didn’t know what the site was for. And a publicly known reason for that website’s existence is because bullies need a place to coordinate harassment.
Having the power to shut that place and therefore behavior down but choosing not to means they were complicit. Deplatforming works, the influx of new bullies shrinks when you make their place less public.
That was unequivocally not a publicly known reason. There were thousands upon thousands of “lolcows” documented in kiwifarms and the sites culture strongly emphasized not “disturbing the grazing lolcows” or “taking things IRL.” That much was obvious for me as somebody who only visited the site a few times a year out of curiosity.
That may be obvious to you. Have you tried asking some of the people documented there what they think about the site's culture, and how well it discourages harassment against them?
I’ve been on the internet long enough to know what harassment is, and I went on Kiwi Farms enough to know the site culture. There will always be people laughing at other people on the internet and I would prefer it stays that way.
The removal in question is an archive. A read-only snapshot cannot be used to coordinate anything. But it can be used to view the documentation they have collected.
Therefore it would seem that this particular removal is motivated by erasing the documentation and performing a cover up.
KF has, at most, 3 deaths debatably attributable to them.
A dozen children died because of the Tide Pod challenge-- and that's just one campaign. We could talk about planking falls, cinnamon poisonings and more if you like. That's before we get to the rise in runaways and child exploitation directly attributable to the "legitimate" platforms. Kids are being solicited on Roblox FFS.
How does a community of bullies deliberately targeting (internet celebrities) have a lower kill count and merit higher priority than a handful of individuals conspicuously encouraging ignorant children to engage in lethal acitivities?
Sure, words can't hurt you, but coordinated harassment campaigns can, and, for the mentally feeble, like CWC, words can coax you, over enough time, to act in your worst interests.
Do you think the parents whose child was on the cover of Nat Geo count as a "public internet persona"? Do you think it's right that people should be forced to go into hiding and avoid something as basic as a magazine interview for fear of becoming a target on these sites?
What you're advocating is complete self-censorship for people with views these websites consider targets.
Being on the cover of a major magazine (with an accompanying article) absolutely makes you a public person, and you can't expect people not to criticize your family for doing something that is still controversial.
Right. Also, we're glossing over the fact that it's not just bullying, but swatting, threatening phone calls, contacting employers to seek termination, email spam, sending unpaid-for pizzas, etc. It's not just words.
So the logical response is going to service providers and business partners to get them deplatformed?
>It's not just bullying
Which is already able to be dealt with as harassment, and is defanged in this case in the judicial system most likely under assumption of risk doctrine as a public figure.
>It's swatting
Already illegal, and has metadata trail through which filing a false report can be dealt with.
>threatening phone calls
Jesus. So block or screen them. Should I go running to the police because I'm harassed by marketers all the time? Furthermore, LE has the metadata to deal with that too, and if they tell you they don't, they're lying.
>contacting employers to seek termination
Twitter does a lot of that too. Sucks when the shoe is on the other foot, doesn't it?
>email spam
Should be a capital offense with death penalty viable, but alas, it's email. Ignore it. Or have fun with it, see how big your collection can get. Start seeing how many times things get repeated. Start doing data analysis on it.
>sending unpaid-for pizzas
Correct response: I'm sorry, but I did not order any. I have a bit of a problem with people online trying to troll me. Here's my contact info, here's a codeword/auth mechanism that can be used to authenticate something came from me in the future. Done. Would recommend actually vetting the establishment in question, and if you know you are bringing this to your locality, be responsible and proactively front-run it by going and talking to local places that deliver to you.
>It's not just words.
It totally is. In point of fact, freedom to communicate without someone else deplatforming you is so important, we call it freedom of speech. Nobody else is forcing the person in question to continue feeding the trolls, and it is a poignant point of fact that if they are disturbed by the observations these people are making, it may be worth taking a moment to reorient around their own actions, and to be a bit more self-conscious on what it is they do that may be setting off others in such a way as it does.
But no, continue to regale us all about how there is no other more reasonable, local, and less harmful way to deal with something like this than blasting the Internet, and encouraging the weaponization of the infrastructure at the core of technological advancement for the last 50 or so years.
