Talking about this like there's an "attack" feels very melodramatic. It seems to me like the non-constructive messages are either shitposts or just curious probes (testing whether certain codepoints make it through, testing whether bad no-no words make it through, testing how various things will render, testing whether the input is sanitized...). There's links to God knows what and some walls of inflammatory language, but I'm not noticing anything that could even remotely cause any harm.
```
The first and most obvious one was a filter of inappropriate words and expressions. I won't explain which ones or how it works inside, for obvious reasons, but it is quite effective. The machinery is in the heuristics.
The second, and most natural, was to cut the maximum number of characters way down. Less room is less ammunition to cover content and less space to hide a link.
```
- you forgot the third fix, disable copy paste inside chat boxes, remove all swear words by replacing them with **
The chat is still there in the corner of this post, with people typing nasty shit into it.
Well have fun delving into programmatic censorship I guess. Or just take it out. If you take the former path and post about trying to fix it here then I'm sure you'll get a lot more free testing of your attempts to enforce civility through regex or whatever. Enjoy learning about the Scunthorpe problem.
Looking at the last few minutes of messages, I get the feeling that all it takes is a handful of motivated (i.e. jobless) malicious individuals to make an entire community or a corner of the Internet look like it has devolved into barbarism.
To get the handful of people to collaborate and hopefully bring their friends. It's currently at 160 and climbing, no longer a handful, so it seems to be working.
From the earliest days of websites allowing text input from users, this has been a thing. The same with any website that allows you to draw things, it will quickly show people drawing crude (in content not skill) objects. This is just human nature. Once people realized you could use these forms as attack vectors, we were off to the races. Now that bots can do it for you, I'd only imagine the time before first bot using the form is in minutes. I have seen all sorts of things suggested as workarounds to mitigate bot form submissions, but eventually, you will be spammed at the least with the feature. Adding something like a bot chat, of course people are going to be abusive to it much more than a simple form field.
For someone to be surprised by this today suggests to me that the person is really really new to managing a website.
Exactly. I remember a website from the late 90s that let anonymous users control an LED sign in some shopping mall in Japan. More often than not there were things like swastikas and penises on the board. This type of behavior is nothing new, unfortunately.
Wait. Is it not against the HN guidelines to regularly submit links to your own sites? Isn't that considered self-promotion and, thus, against the rules? I've got nothing against the author, personally, as I'd do the same, but seeing his submission page, I'd like to know how normal getting in trouble is for this practice or if the guidelines are enforced at all in this sense.
"What I learned" seems to missing what used to be shared to every starting web developer; "If you allow user input on public internet pages, people will put vulgar, racist, hacking attempts and worse there, sometimes constantly over long periods of time"
Almost anyone who had a "guest book" had pre-moderation some way or another, or was a tiny-tiny website with barely any visitors. The second the larger cyberspace ecosystem got a whiff of your user-input-enabled website, the spamming would appear.
I'm mostly curious why anyone would bother with doing attacks like this to some random individuals personal blog, was it one person or multiple people? What was the motivation? Is it just bots crawling the internet to spread hate?
A sad state of affairs, but certainly seems like having a strictly moderated comment section is a better option for a site like this.
I’d suggest one additional heuristic: once your bad word detector has triggered, shut down the chat for 10 minutes. That way waves of assholes don’t monopolize the chat, and when they move on to the next target the functionality returns to your site automatically.
Yeah, shutting down to that specific browser fingerprint would be the better way. Sometimes, being able to recognize a user has it's benefits for good /s
Refreshing pages to establish connection again is semi easy for someone who made a honest mistake (and Could make them more cautious), but if someone wants to trigger web-socket disconnect all the time, it would become cumbersome.
Add something like fail2ban, and you're set enough against passing trollers, but not interesting enough to draw attention from people that like the challenge.
There seems to be one category of technological "attack", which failed - the script injections.
Otherwise:
OK: A guy insults the Tailwinds devs at length
Not OK: Anyone insults that guy briefly
Of course it is his own platform (blog) but that is on another platform (hosting) which could maybe decide they don't want their platform used to attack open source projects.
(I personally have no opinion on Tailwinds except a default negative valence regarding front-end)
You just got yourself a product right there if it works and can market it as Roast me Blog with just trigger filters in your chat to make it healthy roast in place of abuse and highlight top roasters where people could upvote right there in your chat. Negativity turned into positive healthy banter ..
Content moderation and social anonymous behaviour aside, your real-time chat is really poorly implemented UX wise.
I saw the same message appearing more than once, I never saw my own message appear, it has a weird lag/delay feel to it. I didn't really enjoy that experience at all.
Obviously insults and attacks are unreasonable, but I think the tag injection etc should be an expected one if you’re posting on HN right? Like somebody is going to try it for kicks and I wouldn’t say it’s even necessarily malicious
This sounds like the kind of feature that ends up being quick to initially implement but takes up way more engineering time to maintain than everything else combined. The author concedes at the end that most genuine uses of the feature are people just saying hello. So maybe the solution here is to disallow unrestricted input and just have a few buttons with fixed things to say.
While that's undoubtedly true, the level of trolling varies a lot so I don't think it's quite as simple as "if people can troll then they will." There's a crazy level of trolling on 4chan, quite a lot on Reddit or YouTube, and much less on HackerNews (cue troll replies to this post :D ).
I suspect that the owner, or the users, of a site are able steer how much trolling there is by incentivizing good conduct and 'policing'. If people actually value the content they won't troll, or they'll actively keep things tidy using the available tools. If people don't really care about the content (or if they only care about their relatively small corner) then they'll happily trash the bits they don't see value in.
I'll have to be honest here man, yeah no shit. I can't think of any good reason of why you would add something like this. A standard comment section is typically bad enough, and at least you're able to moderate that. An anonymous real-time chat embedded within your own blog? That's just asking for issues.
Most people aren't awful in real life. Seems like a good idea to have a honey pot to just ban everyone in real time that tries to be an awful person on your own blog
My blog posts don't get any replies even when I post them on Hacker News, but everyone wants to communicate with others. I actually think anonymous chat isn't that bad. I think the problem is people who abuse the goodwill of writing. Once you step away from Medium and Substack, you start wondering how to even get comments on your own site.
Ages and ages ago I had a guestbook, forum and chatroom on my site just because I could, just because it was fun. I think only one other person ever used it, though. But the possibility of having a random interaction with someone was what made it interesting.
Now my comments feed is my curated Mastodon account, very much a filter for nonsense. I wouldn't even think of putting a chat up on my site, it would almost entirely be bots, and then trolls. There's literally no point in trying to interact with people online anymore without some kind of curation or moderation, every possible interaction has to be treated as hostile.
> The first and most obvious one was a filter of inappropriate words and expressions
I don't think those countermeasures are working because as of this moment, there's plenty of obvious naughty words being passed through in that chat window.
I was watching porn recently and the particular site I was on had recently added a real time chat widget too! Aside from scam bots trying to lure users off onto telegram, the last few messages were guys talking about Monty Python's The Life of Brian.
The homepage design is nice, and I think anonymous chat is fun too. Even though the programming topic probably has a pretty clear target audience, I'm surprised that kind of hateful chat still shows up.
> What harm could a box that only broadcasts plain, ephemeral text do to me? You're thinking the same thing I was, I'm not crazy, right?
Not right. As soon as I saw the box (when the other article was posted), my immediate thought was that it was distracting, frankly a bit creepy (even just the counter is so), and obviously ripe for abuse. About one second after I had the thought, I saw it happen in real time.
> The goal, I suppose, was twofold: (…) and to make the article look bad in front of the aggregators and networks where it was being shared.
I think you’re reading too much into it. I bet all (or essentially all) the people doing that don’t care one iota about you or how you look to whatever aggregator and network. They’d write those same things on an empty wall if you gave them a can of paint. You gave them an avenue and they used it, simple as that.
Yeah, maximum concurrent clients being 50 is nothing and you're already seeing a large gap. That is the whole point of the BEAM - it scales well with concurrent stuff which is important with web in general and very important for LiveView.
Talking about this like there's an "attack" feels very melodramatic. It seems to me like the non-constructive messages are either shitposts or just curious probes (testing whether certain codepoints make it through, testing whether bad no-no words make it through, testing how various things will render, testing whether the input is sanitized...). There's links to God knows what and some walls of inflammatory language, but I'm not noticing anything that could even remotely cause any harm.
It’s a blog post written by AI. AI makes everything ponderously dramatic.
``` The first and most obvious one was a filter of inappropriate words and expressions. I won't explain which ones or how it works inside, for obvious reasons, but it is quite effective. The machinery is in the heuristics.
The second, and most natural, was to cut the maximum number of characters way down. Less room is less ammunition to cover content and less space to hide a link. ```
- you forgot the third fix, disable copy paste inside chat boxes, remove all swear words by replacing them with **
The chat is still there in the corner of this post, with people typing nasty shit into it.
Well have fun delving into programmatic censorship I guess. Or just take it out. If you take the former path and post about trying to fix it here then I'm sure you'll get a lot more free testing of your attempts to enforce civility through regex or whatever. Enjoy learning about the Scunthorpe problem.
Looking at the last few minutes of messages, I get the feeling that all it takes is a handful of motivated (i.e. jobless) malicious individuals to make an entire community or a corner of the Internet look like it has devolved into barbarism.
By my recollection the devolution started around 30 years ago.
When I checked it out, the chat was a cesspool of racism, zionism, and horny dudes. Occasionally cool fish emojis.
Neuralink will hopefully filter that out client side /s
If your audience is more than a handful of people, what's the purpose of chat?
To get the handful of people to collaborate and hopefully bring their friends. It's currently at 160 and climbing, no longer a handful, so it seems to be working.
What I mean is that I get the purpose of chat when the target audience is a small group of people that might have a reason to talk to each other.
The counter seems broken. It jumps from 100 to 250 then back down to 120. It did that a few times while I was reading.
From the earliest days of websites allowing text input from users, this has been a thing. The same with any website that allows you to draw things, it will quickly show people drawing crude (in content not skill) objects. This is just human nature. Once people realized you could use these forms as attack vectors, we were off to the races. Now that bots can do it for you, I'd only imagine the time before first bot using the form is in minutes. I have seen all sorts of things suggested as workarounds to mitigate bot form submissions, but eventually, you will be spammed at the least with the feature. Adding something like a bot chat, of course people are going to be abusive to it much more than a simple form field.
For someone to be surprised by this today suggests to me that the person is really really new to managing a website.
Exactly. I remember a website from the late 90s that let anonymous users control an LED sign in some shopping mall in Japan. More often than not there were things like swastikas and penises on the board. This type of behavior is nothing new, unfortunately.
Wait. Is it not against the HN guidelines to regularly submit links to your own sites? Isn't that considered self-promotion and, thus, against the rules? I've got nothing against the author, personally, as I'd do the same, but seeing his submission page, I'd like to know how normal getting in trouble is for this practice or if the guidelines are enforced at all in this sense.
"What I learned" seems to missing what used to be shared to every starting web developer; "If you allow user input on public internet pages, people will put vulgar, racist, hacking attempts and worse there, sometimes constantly over long periods of time"
Almost anyone who had a "guest book" had pre-moderation some way or another, or was a tiny-tiny website with barely any visitors. The second the larger cyberspace ecosystem got a whiff of your user-input-enabled website, the spamming would appear.
I'm mostly curious why anyone would bother with doing attacks like this to some random individuals personal blog, was it one person or multiple people? What was the motivation? Is it just bots crawling the internet to spread hate?
A sad state of affairs, but certainly seems like having a strictly moderated comment section is a better option for a site like this.
Bored kids. Bots. And scammers/spammers. On the scale of the internet, there are a lot of all of these.
Same reason any blank wall in a city will soon be covered in graffiti.
I’d suggest one additional heuristic: once your bad word detector has triggered, shut down the chat for 10 minutes. That way waves of assholes don’t monopolize the chat, and when they move on to the next target the functionality returns to your site automatically.
Great way to get someone to create a bot that swears in the chat every 10 minutes and 1 second to permanently shut it down.
But then one asshole can denial of service your chat very easily
Yeah, shutting down to that specific browser fingerprint would be the better way. Sometimes, being able to recognize a user has it's benefits for good /s
Or just drop websocket connection to them.
Refreshing pages to establish connection again is semi easy for someone who made a honest mistake (and Could make them more cautious), but if someone wants to trigger web-socket disconnect all the time, it would become cumbersome.
Add something like fail2ban, and you're set enough against passing trollers, but not interesting enough to draw attention from people that like the challenge.
But don't let them know they've been shut down, use a shadowban.
>> The countermeasures
>> The first and most obvious one was a filter of inappropriate words and expressions.
Wouldn't the first and most obvious one be "drop real-time chat" from a blog?
There seems to be one category of technological "attack", which failed - the script injections.
Otherwise:
OK: A guy insults the Tailwinds devs at length
Not OK: Anyone insults that guy briefly
Of course it is his own platform (blog) but that is on another platform (hosting) which could maybe decide they don't want their platform used to attack open source projects.
(I personally have no opinion on Tailwinds except a default negative valence regarding front-end)
Reminds me of Maria Abramovic Rythym 0 performance art.
1: https://www.thecrimson.com/article/2023/3/30/maria-abramovic...
You just got yourself a product right there if it works and can market it as Roast me Blog with just trigger filters in your chat to make it healthy roast in place of abuse and highlight top roasters where people could upvote right there in your chat. Negativity turned into positive healthy banter ..
Content moderation and social anonymous behaviour aside, your real-time chat is really poorly implemented UX wise.
I saw the same message appearing more than once, I never saw my own message appear, it has a weird lag/delay feel to it. I didn't really enjoy that experience at all.
Obviously insults and attacks are unreasonable, but I think the tag injection etc should be an expected one if you’re posting on HN right? Like somebody is going to try it for kicks and I wouldn’t say it’s even necessarily malicious
The obvious solution is shutting it down it does offer nothing.
Short of that shadow ban everyone everyone sees their own messages, and add a few fake ones every now and then.
Hah, or burn lots of GPU cycles for an LLM-enhanced echo-chamber of one... (For each troll an echo chamber/an LLM troll responder)
There is a well-reviewed study on this from 2004 -
Greater Internet Fuckwad Theory[1]
[1] https://www.penny-arcade.com/comic/2004/03/19/green-blackboa...
Well, you've just painted a big target on your blog. This is the internet, after all.
Well it looked pretty quiet until HN linked to it. So the people writing that terrible stuff are from here...
"Every input is hostile until proven otherwise": you said it yourself. One of those sad things that somebody new learns on the Internet every day.
Next we're going to hear a complaint from Mark Zuckerberg about how people are using Facebook and LLaMa to attack Meta.
SLM language filters might actually be a good application of AI? As well as very sparing regex for bad words.
This sounds like the kind of feature that ends up being quick to initially implement but takes up way more engineering time to maintain than everything else combined. The author concedes at the end that most genuine uses of the feature are people just saying hello. So maybe the solution here is to disallow unrestricted input and just have a few buttons with fixed things to say.
This should come as a surprise to absolutely nobody.
But hey traffic is traffic I guess.
Humans will really take any chance they can to troll, I would definitely get more moderation
While that's undoubtedly true, the level of trolling varies a lot so I don't think it's quite as simple as "if people can troll then they will." There's a crazy level of trolling on 4chan, quite a lot on Reddit or YouTube, and much less on HackerNews (cue troll replies to this post :D ).
I suspect that the owner, or the users, of a site are able steer how much trolling there is by incentivizing good conduct and 'policing'. If people actually value the content they won't troll, or they'll actively keep things tidy using the available tools. If people don't really care about the content (or if they only care about their relatively small corner) then they'll happily trash the bits they don't see value in.
I'll have to be honest here man, yeah no shit. I can't think of any good reason of why you would add something like this. A standard comment section is typically bad enough, and at least you're able to moderate that. An anonymous real-time chat embedded within your own blog? That's just asking for issues.
Most people aren't awful in real life. Seems like a good idea to have a honey pot to just ban everyone in real time that tries to be an awful person on your own blog
My blog posts don't get any replies even when I post them on Hacker News, but everyone wants to communicate with others. I actually think anonymous chat isn't that bad. I think the problem is people who abuse the goodwill of writing. Once you step away from Medium and Substack, you start wondering how to even get comments on your own site.
According to dead internet theory, these might just as well be bots.
According to current internet theory, these are bots.
You are absolutely right.
Well, Django sucks for real-time chat
The social web has become the adversarial web, unfortunately. You have to have a huge squelch knob if you want any meaningful signal now.
Ages and ages ago I had a guestbook, forum and chatroom on my site just because I could, just because it was fun. I think only one other person ever used it, though. But the possibility of having a random interaction with someone was what made it interesting.
Now my comments feed is my curated Mastodon account, very much a filter for nonsense. I wouldn't even think of putting a chat up on my site, it would almost entirely be bots, and then trolls. There's literally no point in trying to interact with people online anymore without some kind of curation or moderation, every possible interaction has to be treated as hostile.
I tried Mastodon for awhile but found the fediverse was almost as bad as the centralized control verse. It sounds like it works for you, though.
> The first and most obvious one was a filter of inappropriate words and expressions
I don't think those countermeasures are working because as of this moment, there's plenty of obvious naughty words being passed through in that chat window.
I was watching porn recently and the particular site I was on had recently added a real time chat widget too! Aside from scam bots trying to lure users off onto telegram, the last few messages were guys talking about Monty Python's The Life of Brian.
It's interesting to hear a story about 'Life' in the very place where 'Life is created.'
What if he filled the chat with bots and is using all of this as stealth marketing?
Well then it would be one product itself - something like i created bots who roast me on my blog ;)
The homepage design is nice, and I think anonymous chat is fun too. Even though the programming topic probably has a pretty clear target audience, I'm surprised that kind of hateful chat still shows up.
lol hate speech (fun speech lbh) drives traffic
> What harm could a box that only broadcasts plain, ephemeral text do to me? You're thinking the same thing I was, I'm not crazy, right?
Not right. As soon as I saw the box (when the other article was posted), my immediate thought was that it was distracting, frankly a bit creepy (even just the counter is so), and obviously ripe for abuse. About one second after I had the thought, I saw it happen in real time.
> The goal, I suppose, was twofold: (…) and to make the article look bad in front of the aggregators and networks where it was being shared.
I think you’re reading too much into it. I bet all (or essentially all) the people doing that don’t care one iota about you or how you look to whatever aggregator and network. They’d write those same things on an empty wall if you gave them a can of paint. You gave them an avenue and they used it, simple as that.
The attempt to run Javascript as the author mentions is pathetic.
You'd be surprised how often it still works.
Yes. The mindset is pathetic.
TL;DR: this is a toy example, so it doesn't require BEAM VM
Hard to take this seriously without the loadtest with the the large number of concurrent users.
For a number of users concurrently reading a long-tail blog post - you will get good results even using DotCom era HTTP polling.
---
See:
Django LiveView vs Phoenix LiveView: a real benchmark
https://en.andros.dev/blog/80134668/django-liveview-vs-phoen...
What is there to take seriously? Sending a ~50 bytes to ~200 listeners over a websocket about once a second is trivial even for a Raspberry Pi.
It pretends to be serious with 5 tables / charts:
Django LiveView vs Phoenix LiveView: a real benchmark
https://en.andros.dev/blog/80134668/django-liveview-vs-phoen...
Yeah, maximum concurrent clients being 50 is nothing and you're already seeing a large gap. That is the whole point of the BEAM - it scales well with concurrent stuff which is important with web in general and very important for LiveView.
So that’s what’s beneath the usual HN tech bro libertarian veneer.
Oh my bad are yall under the impression that HN is not the majority of the traffic right now lmao