If this is the cokamany bullshit Child Safety Act that the AU Gov put in place with the social media ban and age verification shit™ -- I knew this would happens inevitably. The whole thing was stupid as fuck. Not a single one of us that _actually_ work in this industry were consulted to ask whether we _thought_ this was a good idea, no sir. -- Be careful my fellow Aussies, or we end up letting the government build the Great Wall of Australa!
It applies a 'duty of care' to 'online services', which include 'internet carriage services', 'social media', 'relevant electronic services', 'hosting services', 'search engine services', generative AI services, among others. And for encrypted messaging (which would probably come under 'social media' or 'relevant electronic services'), it only excludes 'lawful communications occurring in private solely between consenting adults'.
And the other big thing is it gives approved researchers authority to use sock puppet identities in contravention of any 'agreement, contract or deed', and gives them immunity from civil consequences.
Reliable sources inform me that the government is never going to abuse this act.
I'm sorry, but what I have to say to that is simple this; Parent should teach their children better, teach them virtues, how the world works, protect them from the horrors that exists. Love and nurture them. Not wash their hands and shrug and expect some lousy politicians to protect their children. Geez fuck.
Forgive me for being passionate. But this isn't the way to solve the problem.
Going "entrance forbidden to the libraries and bookstores: they are full of bad texts (and there are people scribbling dumb harmful things on the billboards there)" is not "protecting children".
(With reference to YouTube, chiefly, but not necessarily excluding other repositories. And not to mention the actual intention to block access to encyclopedias.)
I think that depends a lot on the age of the child in question. I absolutely would simply forbid an 8 year old from entering a library or book store with neo nazi writings or aum shinrikyo literature or Ted Kaczynki’s manifesto easily accessible. Trying to provide appropriate context for everything at that age is simply not feasible.
-- You have treated the «8 year old» as a newborn without previous protective formation - so a sad case already;
-- You have defended barring access to all decent libraries and bookstores - which in their hundreds of thousands of volumes do have lots of trash and biased texts. They also contain the good ones.
It's not entrance forbidden to libraries and bookstores though, it's closer to entrance forbidden to bars and casinos. Modern social media is quite low value and quite predatory.
Right now just for you. I clicked on the app and the first thing it displays is a short, of a guy (probably AI) reading a Reddit post about someone's grandpa wanting to be called Bloodfang, overlaid on Subway Surfers. Not the worst content on there but definitely mindless brainrot.
Unbelievable. I do not really know how to reply politely.
I will repeat: go and hunt the wealth that it does contain. It is not a sign of mental development to pick one piece of trash and say that it exhausts the content. See https://news.ycombinator.com/item?id=50140206
It's not that bad. Hard drives are still about $30 a TB, long-form videos worth preserving are around a GB each in decent quality, so they cost about $0.03 each to preserve in decent quality. It's not free but it's not too expensive either. $0.06 for redundancy, $0.04 if you use erasure coding (e.g. RAID-6).
Reduce that by a factor of 10 if you accept noticeable visual artifacting while still conveying the information - that is, if you believe it's not necessary to preserve every pixel as long as the information content remains. If you're focused on quantity of preservation over quality, you should do this.
Again, you do not seem to be appreciating the context. You are suggesting that cultural preservation and accessibility, that should be a goal of States (good luck nowadays about their goals), and is today in one of its most important cone (video material) in the hands of an enterprise, should be replicated (in its important subsets) by private citizens.
You are forgetting, past scale, legal details (the MIT and Stanford etc., and even the holders of old material, chose to publish their content on YT) and systemic details (Lesics and Asianometry and Alpha Phoenix and Sabine Hossenfelder and Steve Mould and 3blue1brown and Thoisoi and the very many others who are producing content specifically for the platform can do it financially because of visiting users on YT - not on backups), and the absurdity.
May I ask what the motivation behind KilllSwitch is? Out of curiosity, I decided to test your claims and asked Claude to build a program in Go and Claude Skill that is able to decompile (from) and compile (to) KS programs along with a Claude Skill that doesn't trip it up. It all works™ -- All your example programs.
So, whilst I _think_ you've built a very clever interpreter, I don't think you can get around the fact that someone (like me) could just build around the noise you've created anyway.
The motivation is to build a language that is ideally impossible for LLMs to read/write/understand, or at the very least as difficult as possible. It's cool that you've given it a try, but if you feed Claude the KillSwitch source code it can feasibly just analyse the Rust interpreter source code and likely get around understanding the language itself.
I'm much more interested in the benchmarks[1] I have built which do show LLMs not being able to gain full ability to understand it. The benchmarks only give the model access to the interpreter binary and the spec, nothing else and no internet access. I'd also be interested in other's ideas of how to make the language even more resistant to LLMs/coding agents.
I think you missed my point. What I'm saying is that your concern over building a language that LLM(s) initially can't read or understand, whilst is very admirable, is futile IMO. Why? Because you're not trying to protect your code from LLM(s) at all, there are humans behind these tools. e.g: me.
I just demonstrated to you that no amount of obfuscation is going to protect your work from an LLM -- Not whilst there's still humans behind the tools driving them.
I think this language is fascinating, but immediately reached a similar conclusion: it might trip up an LLM consuming the text directly but once an interpreter/compiler/decompiler has been written for the language it's trivial for an agent to consume these programs and produce them.
Now on the other hand, as a class of techniques - "anti-LLM languages" - it might be a useful method of emitting code that some agents with the correct tools can consume and others cannot, by design. Evolving such techniques with differing specifications might make for an interesting line of investigation.
Indeed, and an interpreter has been written[1]. As part of the benchmarks, in some cases the models receive a binary of the interpreter and indeed some of them attempt to reverse engineer it. So that already helps the models a bunch.
I do think this will also eventually make it into the training data, but that will in itself be interesting to see. It's also possible for KillSwitch to evolve and gain/remove features over time to constantly evade the LLMs.
Because that's not the point. Before you could waste, without kidding, hours in meetings discussing css alignments, flows that didn't make sense but the upper levels were pretty serious about it - "Our QA and test groups say this is of the utmost importance", even when the QA and test groups were the same, 3 or 4 people that insisted on it.
I think most people, but at least I, could always deliver things inside schedules (not even set by myself) without a problem, even with those "adjustments" and changes of mind, and doubly so if I could just have the same leeway that now seems to be in place.
Nonetheless, if there's going to be a premium it has to be discernible in any particular way because software for business is always considered a cost. But what really shows this industry is half a joke, is that now there's no issue in giving 3rd party agents full access to inner code and logic, many times credentials too because these systems are being deployed in CIs. And because there's really no enforcement or punishment for leaks, or endangering users in most areas there's no reason for a premium. There's people pushing thousands of lines / min (and I think even the most professional people out there can't honestly keep up with it) although that isn't in itself meaning the overall security is worse than before, it just means that for the most part it's now un-auditable in any meaningful way.
I do prefer OSS over proprietary software because to me, at least prior to GenAI, OSS was the label that was "software made by Humans" where proprietary was "made by corporations". It didn't make OSS win commercially, of course, and I personally don't think that human-made art will win commercially.
Because code is not art and the corporate culture doesn't care about artisan code, neither do the users. As long as the software sells, they are happy.
In fact executives demand the opposite: token max to get the features out of the door faster, all in the same of closing that one last sale.
The software part is quickly being solved but I honestly believe the more creative aspects will still see large demand. LLMs have a certain stink to their UI, being creative is a hard skill, which AI has yet to master.
Most of what gets called creative isn't creative at all. Often it means "Produce a content product which approximately matches commercial genre expectations with just enough of a twist to stand out a little."
A lot of people can do that. A lot more people could do it if they tried to.
People who are genuinely out of the box creatives, with a unique blend of original vision, technical skill, emotional and psychological insight, and genre-changing taste, are not common.
For so-called "AI" to have creativity, we'd first have to work out what that even means for ourselves right? What is creativity in the first place? I'm unconvinced it's just turning up the "temperature" or fiddling with some pertubing parameters.
EDIT: And AI is also pretty terrible at being consistent too, you have to work quite hard to get a level of "consistency" that you're happy with.
There will always be problems to solve with building software, regardless of the way it is written.
From code to no-code to vibe-code, to even more code. What we are seeing here is more of the same being built.
An example of this is "sandboxes". Everyone was writing "sandboxes" until it there was 1 vulnerability needed to break them and we will keep seeing this with developers vibe-coding infrastructure.
Does IRCv3 need new leadership to continue to drive it forward? Over at Parley/Ahoy (https://git.mills.io/prologic/parleyhttps://git.mills.io/fastidious/ahoy) we've managed to create a federated decentralized IRC chat experience with a Discord/Slack-esque feel but true to both IRC / IRCv3 (Ahoy being a fully featured IRCv3 client).
Rather than claim a Discord-esque feel, you should point out the features you're referring to explicitly. Discord is infamous for its absolutely awful UX and hostility to open source ecosystem. I'm sure you're referring to neither of those aspects, but consider that those reading your reference might be entirely unfamiliar with the (presumably) positive aspects that you're actually imitating.
Infamous for it's bad UX? This thread is genuinely the first time I've ever heard this, all of my extended circles were overjoyed when Discord came onto the scene and displaced the patchy, hodgepodge UI's of Mumble, Ventrilo, IRC, Skype simultaneously. Both from a feature and a visual perspective it was a breath of fresh air for the programmers, gamers, and casuals alike for everyone I know. Maybe the only thing that still survives alongside it for the folks I know as a preferred communication channel is sporadic Whatsapp among the EU guys and iOS group chats for the Americans locked into the Apple ecosystem.
What about it makes it terrible UX for you? Feature set seems to get better every year as far as I'm concerned, inclusion of ads aside.
My single biggest complaint, other than it being an Electron app, is their insistence on changing the notification sounds seasonally. I do not want surprise Halloween music when I'm trying to start a video call with someone, nor do I want it themed. You can offer it, sure - I get that it's free, and they're trying to keep users on their platform - but it's just grating to me.
I love Mumble and IRC. You could put a gun to my head and I still wouldn't use Discord. When all these proprietary chats are dead and the kids have moved on to something else, Mumble and IRC will still work. They are eternal. Anything proprietary and centralized can't compete with that. IRC outlived AIM and MSN and so on.
my latest annoyance is the way they keep making the gift nitro button bigger and more likely to be where you tap when trying to send a message. I don't know what their PMs are smoking to think that gifting nitro is something people do so often that there needs to be a permanent flashing button on screen to do it.
To be honest, I don't think most people's first interpretation of someone's claim to a "Discord-esque feel" when describing the UX of their open-source IRC client is that it's therefore riddled with dark patterns to sell an upgraded version of the service. I think it's probably more likely that they were referring to the general layout (eg: having circular server icons as a vertical list on the left hand side). I'm guessing that that's also why they mentioned Slack, since that has the same layout too.
It's laden with ads and microtransactions, but if you remove those (shady scripts abound for those who don't want to do it themselves), the UX is one of the best on the market. There's a reason so many alternative chat clients are basically Discord clones without the microtransactions (for now).
Its open source hostility doesn't really change how well it works. You can copy the good parts without the awful parts, like so many people and companies are already doing.
Laden with ads is an overstatement considering Quests are tucked away in their own menu.
The problem with those many alternate chat clients is that they either require self-hosting, making them a nonstarter for probably 95% or more of Discord's userbase, or their lack of monetisation makes them non-viable long term.
You are right. To be perfectly honest, the only thing I took from Discord at all (it's UI is generally awful yes, I agree!) is the support for multiple servers. But now that I think about it some more, that's not really even a unique feature.
Anyway, you are right. But that's more about Ahoy and less about Parley. Parley was just answering the question of "What if IRC was federated, and what if it borrowed some ideas from Salty.im -- DNS discovery + WKID".
Wasn't IRC always federated? There were different networks, but each network was a federation of independent servers.owmed by independent operators as far as I understand...
> A federated network is a system where independent networks, servers, or organizations agree to follow common standards so they can talk to each other without giving up control to a single central company.
In traditional IRC networks, you end up forming a spanning tree of servers that belong to a common name, "the network". However control remains in the hands of the operators of that network.
The "control" was basically permission to connect (and thus share messages/users/etc) between multiple servers. So I see your point, you couldn't just randomly connect your server to someone else's server in quite the same way email worked, but at the same time the servers could be easily owned by different people running different software.
> the only thing I took from Discord at all (it's UI is generally awful yes, I agree!) is the support for multiple servers.
To be sure, Discord "servers" are not servers in the sense the word is used in virtually every other context. They are more like group chats that share a hosting location. If anything the hard part for Discord is making them seem disparate enough that someone might rent their "own server".
...But one of these things is not like the others.
"Anyone can start a server" is only true when you're lying about what a "server" is. (Or redefining it to the point of usefulness.)
A Discord "server" is just a defined chat namespace. All the Discord chat "servers" live together on the same actual load-balanced VMs or whatever on the physical servers they run.
It's honestly the thing that's annoyed me most about Discord from the very beginning, long before they introduced the ads and the microtransactions and the mandatory doxxing.
At least 4 people I know. It's super easy to put up an instance of ergo now. I was using one for LAN communication during internet outages for a while.
IRC sure hasn't! Well, the IRC most people are using, anyway. You can work around these by using things like bouncers and third party HTTP hosts and assuming everyone else has a client that inlines media.
Multimedia transfer is still a practical impossibility, for better or for worse.
Mobile IRC works, sort of, if you don't mind spamming the channel every time you lose connectivity when you drive into a tunnel or when your phone goes to deep sleep. Again, you can set up a bouncer for that.
I welcome the new changes in IRCv3 and hope that one day soon all IRC users will be able to use the features everyone else has been using for fifteen years or so.
IRCv3 isn't a corporate body, its a semi-formal group of IRC developers. There's nothing prohibiting you from proposing protocol extensions if they are within the scope of the working group.
parley and ahoy both look quite interesting!
Retrofitting irc into a real federated protocol is cool.
Also like the idea of picking up discords ui/ux. How did
you manage to get avatars for ahoy? Is there a ircv3 protocol
i missed?
It's part of the IRCv3 Metadata spec: https://ircv3.net/specs/extensions/metadata -- It just isn't very well documented or prominently displayed on the spec's page(s) I think. Took me a while to find it too.
Have you considered not serving ads on your website and making it more accessible to discourage those from wanting to tip it off and strip the ads? I can tell you from honest experience, if I were a user of your software/site (which I'm not) and it had ads, I'd be dropping them on the floor pronto (probably with AdGuard). I hate ads, I despise what we've done with the Web/Internet with all shit Advertising and Tracking shit™
I don’t like ads either. So I just paid for a local image editor and use that one instead. The developer of the product gets paid, I get a great ad-free image editor.
reply