It is difficult to figure out the incentives here. Why would anyone want to pull data from LWN (or any other site) at a rate which would cause a DDOS like attack?
If I run a big data hungry AI lab consuming training data at 100Gb/s it's much much easier to scrape 10,000 sites at 10Mb/s than DDOS a smaller number of sites with more traffic. Of course the big labs want this data but why would they risk the reputational damage of overloading popular sites in order to pull it in an hour instead of a day or two?
Probably someone who has more access to bandwidth than sense who either rate limited themselves incorrectly (for instance, they forgot to configure it in their production environment and the default value was 0) or simply didn't even think to do so.
It would be funny if someone vibe coded scraping infrastructure and tested it at a small scale, didn't see a problem, and so turned the throttle all the way up.
I don't think that most of them are from big-name companies. I run a personal web site that has been periodically overwhelmed by scrapers, prompting me to update my robots.txt with more disallows.
The only big AI company I recognized by name was OpenAI's GPTBot. Most of them are from small companies that I'm only hearing of for the first time when I look at their user agents in the Apache logs. Probably the shadiest organizations aren't even identifying their requests with a unique user agent.
As for why a lot of dumb bots are interested in my web pages now, when they're already available through Common Crawl, I don't know.
Maybe someone is putting out public “scraper lists” that small companies or even individuals can use to find potentially useful targets, perhaps with some common scraper tool they are using? That could explain it? I am also mystified by this.
I'm not at all sure alibaba or tencent would actually want to DDOS LWN or any other popular website.
They may face less reputational damage than say Google or OpenAI would but I expect LWN has Chinese readers who would look dimly on this sort of thing. Some of those readers probably work for Alibaba and Tencent.
I'm not necessarily saying they wouldn't do it if there was some incentive to do so but I don't see the upside for them.
I worked on an extremely niche project revolving around an old DOS game. Code I worked on is often pretty much the only reference for some things.
It's trivially easy to get claude to scrape that and regurgitate it under any requested licence (some variable names changes, but exactly the same structure - though it got one of the lookup tables wrong, which is one of the few things you could argue aren't copyrighted there).
It'll even cheerfully tell you it's fetching the repository while "thinking".
If I referenced copywritten code we didn't have the license for (as is the case for copyleft licenses if you don't follow the restrictions) while writing "my own" I'd be fired pretty quick.
> AI allows companies to resell open source code as if they wrote it themselves doing an end run around all license terms. This is a major problem.
Has it been adjudicated that AI use actually allows that? That's definitely what the AI bros want (and will loudly assert), but that doesn't mean it's true.
From the creators of easy money laundering (crypto bros), we now bring you easy money laundering 2: intellectual property laundering, coming to a theatre near you soon!
>From the creators of easy money laundering (crypto bros),
Is there even any evidence that "crypto bros" and "AI bros" are even the same set of people other than being vaguely "tech" and hated by HN? At best you have someone like Altman who founded openai and had a crypto project (worldcoin), but the latter was approximately used by nobody. What about everyone else? Did Ilya Sutskever have a shitcoin a few years ago? Maybe Changpeng Zhao has an AI lab?
That was a biometric surveillance project disguised as a crypto project.
> Is there even any evidence that "crypto bros" and "AI bros" are even the same set of people
No, the "AI" people are far worse. I always had a choice to /not/ use crypto. The "AI" people want to hamfistedly shove their flawed investment into every product under the sun.
I am starting to think these are not just AI scrapers blindly seeking out data. All kinds of FOSS sites including low volume forums and blogs have been under this kind of persistent pressure for a while now. Given the cost involved in maintaining this kind of widespread constant scraping, the economics don’t seem to line up. Surely even big budget projects would adjust their scraping rates based on how many changes they see on a given site. At scale this could save a lot of money and would reduce the chance of blocking.
I haven’t heard of the same attacks facing (for instance) niche hobby communities. Does anyone know if those sites are facing the same scale of attacks?
Is there any chance that this is a deniable attack intended to disrupt the tech industry, or even the FOSS community in particular, with training data gathered as a side benefit? I’m just struggling to understand how the economics can work here.
How many of these scrapers are written by AI by data-science folks who don't remotely care how often they're hitting the sites, and is data they wouldn't even think to give or ask the LLM about?
But does that explain all of the various scrapers doing the same thing across the same set of sites? And again, the sheer bandwidth and CPU time involved should eventually bother the bean counters.
I did think of a couple of possibilities:
- Someone has a software package or list of sites out there that people are using instead of building their own scrapers, so everyone hits the same targets with the same pattern.
- There are a bunch of companies chasing a (real or hoped for) “scraped data” market, perhaps overseas where overhead is lower, and there’s enough excess AI funding sloshing around that they able to scrape everything mindlessly for now. If this is the case then the problem should fix itself as funding gets tighter.
I've had luck blocking scrapers by overwriting JavaScript methods
" a.getElementsByTagName = function (...args) {//Clear page content}"
One can also hide components inside Shadow DOM to make it harder to scrape.
However, these methods will interfere with automated testing tools such as Playwright and Selenium. Also, search engine indexing is likely to be affected.
"It is a DDOS attack involving tens of thousands of addresses"
It is amazing just how distributed some of these things are. Even on the small sites that I help host we see these types of attacks from very large numbers of diverse IPs. I'd love to know how these are being run.
There are plenty of providers selling "residential proxies", distributing your crawler traffic through thousands of residential IPs. BrightData is probably the biggest, but its a big and growing market.
And if you don't care about the "residential" part you can get proxies with data center IPs for much cheaper from the same providers. But those are easily blocked
In the most charitable case it's some "AI" companies with an X/Y problem. They want training data so they vibe code some naive scraper (requests is all you need!) and don't ever think to ask if maybe there's some sort of common repository of web crawls, a CommonCrawl if you will.
They don't really need to scrape training data as CommonCrawl or other content archives would be fine for training data. They don't think/know to ask what they really want: training data.
In the least charitable interpretation it's anti-social assholes that have no concept or care about negative externalities that write awful naive scrapers.
Is it still ongoing? The thread appears to be over 24 hours old and as a quick test I had no issue loading the main page (which is as snappy and responsive as expected from a low-bandwidth site like LWN).
If it's a conspiracy, it would be one where the Minimum Viable Conspirator Count is 1 (inclusive of one's own self).
In that case, by that rubric literally anything that you conspire with yourself to accomplish (buying next week's groceries, making a turkey sandwich...) would also be a conspiracy.
Umm... what data? That's a very old newsletter-like site. Everything that's public on it has been long scraped and parsed by whoever needed it. There's 0 valuable data there for "parasites" to parasite off of.
I also don't get the comments on the linked social site. IIUC the users posting there are somehow involved with kernel work, right? So they should know a thing or two about technical stuff? How / why are they so convinced that the big bad AI baddies are scraping them, and not some miss-configured thing that someone or another built? Is this their first time? Again, there's nothing there that hasn't been indexed dozens of times already. And... sorry to say it, but neither newsletters nor the 1-3 comments on each article are exactly "prime data" for any kind of training.
These people have gone full tinfoil hat and spewing hate isn't doing them any favours.
Your nonsense about LWN being a "newsletter" and having "zero valuable data" isn't doing you any favors. It is the prime source of information about Linux kernel development, and Linux development in general.
"AI" cancer scraping the same thing over and over and over again is not news for anybody even with a cursory interest in this subject. They've been doing it for years.
Again, the site is so old that anything worth while is already in cc or any number of crawls. I am not saying they weren't scraped. I'm saying they likely weren't scraped by the bad AI people. And certainly not by AI companies trying to limit others from accessing that data (as the person who I replied to stated).
I’m going to presume good faith rather than trolling. Some questions for you:
1. Coding assistants have emerged as as one of the primary commercial opportunities for AI models. As GP pointed out, LWN is the primary discussion for kernel development. If you were gathering training data for a model, and coding assistance is one of your goals, and you know of a primary sources of open source development expertise, would you:
(a) ignore it because it’s in a quaint old format, or
(b) slurp up as much as you can?
2. If you’d previously slurped it up, and are now collating data for a new training run, and you know it’s an active mailing list that will have new content since you last crawled it, would you:
(a) carefully and respectfully leave it be, because you still get benefit from the previous content even though there’s now more and it’s up to date, or
(b) hoover up every last drop because anything you can do to get an edge over your competitors means you get your brief moment of glory in the benchmarks when you release?
I'm curious how they concluded this was done to scrape for AI training. If the traffic was easily distinguishable from regular users, they would be able to firewall it. If it was not, then how can they be sure it wasn't just a regular old malicious DDOS? Happens way more often than you might think. Sometimes a poorly-managed botnet can even misfire.
Why would anyone ever DDOS them? They’ve been around for about three decades now, I don’t know if they’ve ever had a DDOS attack before the AI crawling started.
It is difficult to figure out the incentives here. Why would anyone want to pull data from LWN (or any other site) at a rate which would cause a DDOS like attack?
If I run a big data hungry AI lab consuming training data at 100Gb/s it's much much easier to scrape 10,000 sites at 10Mb/s than DDOS a smaller number of sites with more traffic. Of course the big labs want this data but why would they risk the reputational damage of overloading popular sites in order to pull it in an hour instead of a day or two?
It would be funny if someone vibe coded scraping infrastructure and tested it at a small scale, didn't see a problem, and so turned the throttle all the way up.
The only big AI company I recognized by name was OpenAI's GPTBot. Most of them are from small companies that I'm only hearing of for the first time when I look at their user agents in the Apache logs. Probably the shadiest organizations aren't even identifying their requests with a unique user agent.
As for why a lot of dumb bots are interested in my web pages now, when they're already available through Common Crawl, I don't know.
They may face less reputational damage than say Google or OpenAI would but I expect LWN has Chinese readers who would look dimly on this sort of thing. Some of those readers probably work for Alibaba and Tencent.
I'm not necessarily saying they wouldn't do it if there was some incentive to do so but I don't see the upside for them.
Of course they're not going to stop at just code. They need all the rest of it as well.
It's trivially easy to get claude to scrape that and regurgitate it under any requested licence (some variable names changes, but exactly the same structure - though it got one of the lookup tables wrong, which is one of the few things you could argue aren't copyrighted there).
It'll even cheerfully tell you it's fetching the repository while "thinking".
If I referenced copywritten code we didn't have the license for (as is the case for copyleft licenses if you don't follow the restrictions) while writing "my own" I'd be fired pretty quick.
Has it been adjudicated that AI use actually allows that? That's definitely what the AI bros want (and will loudly assert), but that doesn't mean it's true.
Is there even any evidence that "crypto bros" and "AI bros" are even the same set of people other than being vaguely "tech" and hated by HN? At best you have someone like Altman who founded openai and had a crypto project (worldcoin), but the latter was approximately used by nobody. What about everyone else? Did Ilya Sutskever have a shitcoin a few years ago? Maybe Changpeng Zhao has an AI lab?
That was a biometric surveillance project disguised as a crypto project.
> Is there even any evidence that "crypto bros" and "AI bros" are even the same set of people
No, the "AI" people are far worse. I always had a choice to /not/ use crypto. The "AI" people want to hamfistedly shove their flawed investment into every product under the sun.
I haven’t heard of the same attacks facing (for instance) niche hobby communities. Does anyone know if those sites are facing the same scale of attacks?
Is there any chance that this is a deniable attack intended to disrupt the tech industry, or even the FOSS community in particular, with training data gathered as a side benefit? I’m just struggling to understand how the economics can work here.
I did think of a couple of possibilities:
- Someone has a software package or list of sites out there that people are using instead of building their own scrapers, so everyone hits the same targets with the same pattern.
- There are a bunch of companies chasing a (real or hoped for) “scraped data” market, perhaps overseas where overhead is lower, and there’s enough excess AI funding sloshing around that they able to scrape everything mindlessly for now. If this is the case then the problem should fix itself as funding gets tighter.
" a.getElementsByTagName = function (...args) {//Clear page content}"
One can also hide components inside Shadow DOM to make it harder to scrape.
However, these methods will interfere with automated testing tools such as Playwright and Selenium. Also, search engine indexing is likely to be affected.
And if you don't care about the "residential" part you can get proxies with data center IPs for much cheaper from the same providers. But those are easily blocked
They don't really need to scrape training data as CommonCrawl or other content archives would be fine for training data. They don't think/know to ask what they really want: training data.
In the least charitable interpretation it's anti-social assholes that have no concept or care about negative externalities that write awful naive scrapers.
big tech incentivised to ddos... what a world they've built
In that case, by that rubric literally anything that you conspire with yourself to accomplish (buying next week's groceries, making a turkey sandwich...) would also be a conspiracy.
I also don't get the comments on the linked social site. IIUC the users posting there are somehow involved with kernel work, right? So they should know a thing or two about technical stuff? How / why are they so convinced that the big bad AI baddies are scraping them, and not some miss-configured thing that someone or another built? Is this their first time? Again, there's nothing there that hasn't been indexed dozens of times already. And... sorry to say it, but neither newsletters nor the 1-3 comments on each article are exactly "prime data" for any kind of training.
These people have gone full tinfoil hat and spewing hate isn't doing them any favours.
https://lwn.net/Articles/1008897
Your nonsense about LWN being a "newsletter" and having "zero valuable data" isn't doing you any favors. It is the prime source of information about Linux kernel development, and Linux development in general.
"AI" cancer scraping the same thing over and over and over again is not news for anybody even with a cursory interest in this subject. They've been doing it for years.
I mean...
Again, the site is so old that anything worth while is already in cc or any number of crawls. I am not saying they weren't scraped. I'm saying they likely weren't scraped by the bad AI people. And certainly not by AI companies trying to limit others from accessing that data (as the person who I replied to stated).
1. Coding assistants have emerged as as one of the primary commercial opportunities for AI models. As GP pointed out, LWN is the primary discussion for kernel development. If you were gathering training data for a model, and coding assistance is one of your goals, and you know of a primary sources of open source development expertise, would you:
2. If you’d previously slurped it up, and are now collating data for a new training run, and you know it’s an active mailing list that will have new content since you last crawled it, would you:Edit: Fabian2k was ten seconds ahead. Damn!