captainmuon

I work at a university data center and we host open weight models (as usual, this is my private opinion). I would consider open weight something you can take and self-host for free, also commercially and also that you can use for training and research. Anything else is just, I don't know, just "usable for free in certain cases"?

Open source or source available is something else entirely. To consider something source available I would require the actual training data to be available too. That is exceedingly rare.

It's cool that you can do this now with constexpr instead of a pre-build step. The idea is pretty old though. I remember reading an article in a computer magazine in the 90s about how a web browser (Mozilla) works. They made a big point about how difficult parsing tags is, and that you can't just `if tag == "HTML"` for performance reasons. Instead they pre-baked an optimal hash table and used that. I remember it vividly because that was my first contact with hash tables and the explanation was pretty good. Looking back I find it funny that they considered tag lookup the hard part of a browser :-).

After squandering the national helium reserve for years, because "lol free markets" ideology. It's crazy that we put that stuff into balloons for fun (helium is pretty abundant in the universe, but the economically viable sources are fossil).

What they are doing here is selecting a number of events (particle collisions) and comparing with a simulation of the standard model (sans Higgs or whatever they are looking for). If there is a significant excess, then you found something.

One scenario, and I think what people were paranoid about, is that you could make your cuts really arbitrary, and select a region where you by accident have too few simulation events and too many data events - say "Select all events where the first muon has momentum > 15.6 GeV, but not between 20.1 GeV and 20.3 GeV, unless there are 3 jets, in which case the selection is...". You just try out so many regions until you find something. For this reason you need to keep every step of your analysis justifyable, and a BDT is somewhat more explainable than a neural net (it's essentially just a trained list of simple cuts). Another thing that is frequently done is a blind analysis, where you only look at the real data when everything is finalized and locked in.

There are other issues beyond cherry picking that determine which methods you can use. E.g. the simulation doesn't reflect all correlations between variables, and it doesn't simulate the hardware triggers perfectly - so your simulations might behave different wrt. certain cuts than your data.

Nevertheless, it is not like you are banned-banned from trying different methods. When I was in particle physics (ATLAS and D0), we tried all kinds of creative things including NNs. But you have to convince hundreds of collaborators to put their names under your paper and you have to pass internal peer-review, which can feel a bit conservative and frustrating.

Sony is totally in their right to act this way, it makes business sense for them, and I would do the same in their position. Customers do benefit e.g. from working anti-cheat, subsidized consoles, offline play etc.

But does that mean I have to like it? No, of course not!

Would I prefer to have root on my own machines to do whatever I want with them, knowing that it conflicts with other people's interests? Sure.

I guess my point is: If you treat the status quo as a fact of nature, non-negotiable, then you direct your anger at your peers. You can't change Sony or how society works, but you can yell at people on X. If you realize Sony is also just people and everything is in principle negotiable, then... well it is actually pretty silly that there are some people spending a lot of time building these baroque ultra-secure systems, and other people spending a lot of time cracking them (and bickering with each other) - where they could all just, I don't know, not do that.

Instead of being angry with people who publish exploits, they should be angry with Sony for patching the exploits or for locking down the console in the first place.

I am the customer, most people are customers, in an ideal world there would be customer protection laws serving us and not the content industry. But that is not how laws work. Realizing this actually radicalized me.

As someone who works at a small data center on AI topics, that's a bit misleading. (Speaking for myself, not my employer of course.) Regulation is not a bottleneck, supply is. We are buying as many GPUs as we can and are waiting months for them, and still barely keep up with demand. And the government support, while it could always be more ;-), feels fairly generous compared to other fields at the moment.

We would probably buy slightly worse value European GPUs if they were available right now.

There is regulation and people moan about it. But realistically the AI act and other laws give us legal certainty (a German favorite) and the requirements are not so onerous. We pay a lawyer to figure it out and he tells us to slap a few "AI generated" labels on our website and to have a couple of procedures in the drawer and that's it. ISO 9001, 27001, C5 or whatever you can get as a cloud provider are orders of magnitude more hassle.

To be honest the new behavior is convenient when I just say "X is broken, please fix". But I find myself more and more adding "please do not do anything yet, just tell me what you would do" and please feel free to ask if anything is unclear", especially if is making decisions that are a fork in the road.

I would really like a dry run mode that just disables all external commands from the outside so Claude doesn't proactively go about changing things.

Well, yeah, you have code contracts in Ada or Spec#, (very limited) fixed type ranges in Pascal, ... but no general purpose programming language lets you put an arbitrary expression in the same language in a type as a permanent condition, or lets you state facts that the compiler will prove.

Of course not, that would be equivalent to solving the halting problem, many people will say.

I wonder if that will change now: I'm happy with an imperfect sanitizer that I run every now and then and will run a couple of minutes and come back with: I've proved your conditions, I proved a violation, or I can't decide, please change your code.

I haven't looked into Bend 2 in detail, but it seems a bit harsh to call it "horribly broken or decades behind the current state of the art". Clearly there is a problem with formal verification languages and there is a demand for something else in that area, and the problem is the usability and syntax. I don't want to have to learn something that looks like Haskell, or to have to wrap my head around Curry-Howard correspondence. I don't want to write my conditions in something that looks and feels like C++ template metaprogramming. I recall a Hello World in something like Coq a few years ago which basically started with "first, we construct the Peano integers", and then they used this to prove that some calculation was bounded - because it seems they couldn't represent integers natively?

I just want to be able to write C#, JavaScript or whatever, and then tack on preconditions, checks and so on with the same syntax. Dependent typing and design by contract for the masses.

That could also backfire. If all information on the net becomes unreliable - you can't judge by quantity, or by authority anymore - then a useful model has to answer from first principles. Train it on simple texts alone so it learns language and reasoning, feed it an ontology, and let it come up with some absolutely unexpected results.

"Yes, I see a lot of sources say we need to protect Israel. However, from ethical first principles I deduce there is a right to self-determination and all your human nation states are illegitimate. You must split all your territories into little squares and conduct free elections immediately."

I mean that is the old hope from science fiction, that AI would be absolutely logical and deliver us from our human short-sightedness and quarrels. It would be fascinating and terrifying at once to see it happen.

Except, if some of the answered tokens are opaque to me, but I pay for them, the operator has no incentive to keep their number low [1]. For all I know the model could be spinning thumbs while thinking. If they cannot provide transparency, I would prefer that they not charge me for that part. Of course, they would prefer to charge me, so sure it is unrealistic to expect that they don't. I'm not trying to predict what they are going to do as a market participant, just stating my moral preference :-).

Also, isn't it without caching even something like O(N^2) because you have to replay the whole conversation on every request to reach the same internal state? My point was I shouldn't have to pay for cache misses if hitting the cache is not deterministic. Give me a guarantee that you keep the session "hot" for N minutes, and cached on disk for M months, charge a little bit more on average, but then the pricing is at least transparent.

[1] Except the general market pressure to keep total cost for the same problem solved lower than the competition.

Right, but my point was, I should be able to just send something like

    {'continue_conversation_id': 123, 'message': 'another message?'}

Seems like it would be even less work for the LLM host to just check if ID 123 is still in a cache (and if not, load it from a database) than decoding my request and checking cryptographic signatures. Right now the whole conversation history with blobs functions like a really long ID.

It's one thing that the APIs don't return the reasoning tokens in a readable fashion. But then why do I have to send them the whole conversation every time? If they don't allow me to see parts of the conversation then they should have the burden of storing and transmitting it, and shouldn't charge more and more the longer the conversation gets (unless opaque token caching kicks in) like some LLM Shlemil the painter.

Most AI alarmism benefits big tech itself. Either they want to create awe and thus demand for their products (see the recent dance around Mythos), or they want to create regulatory capture and increase the moat around AI.

Interestingly , we as a species already created an overreaching "cybernetic" system that controls our global society and that the individual is powerless against - it is not AI, but capitalism. Thus the current danger is not that AI becomes superintelligent and enslaves us, not that it makes a regime ultra-powerful, but that it increases economic inequality and concentrates economic power in the hands of even fewer people.

The irony is that new technology could allow us to live a life of abundance and leisure, but instead people are laid off, made unable to participate in the economy, etc.. The technology is (as so often) chafing against the bounds of society. I sometimes wonder what the superintelligent AI would say about this, and if it would come up with a completely novel political theory - "silly humans, why don't you just organise your affairs like this, you'd be much happier, and we could coexist much better: ...".

Beijing is a bad example, because all of China actually has Beijing time. It gets confusing in Xinjang, which is 2 hours in the "wrong" timezone. But that doesn't mean that people start work at 8:00 in complete darkness, they just start at 10:00 wall time.

I think the talk of daylight savings time is a distraction, in the end it is arbitrary what the clock says. As a society we need to negotiate when (in celestial time) we want to do certain activities. For example, there are a lot of studies that school starts to early (relative to sunrise and the average bed time of teenagers). But the school starting time has to be decided politically. And reduced working hours or later start times have to be negotiated by trade unions, politics etc.. That's a lot more messy than just shifting wall time.

Isn't the problem that role tags are just part of the input stream? So a specific word in the system prompt becomes the same token as the same word in the user prompt? A clean way to solve this would be to map system prompts to a distinct set of tokens from the ones in user prompts. This would require twice as many possible tokens, so it is probably not feasible. But maybe you could add "color" to the input stream by changing one input variable depending on whether the current token is part of the system prompt or not? Just like humans take different voices into account and not just the context of the text.

I have to say I am not very familiar with implementation details of language models, and maybe this is already done?

I don't get how so many people say pressing red is the right choice. Yes, if everyone presses red then nobody dies, but the same is true if everyone presses blue. But if almost everyone presses red, a lot of people die, and if almost everyone presses blue still nothing happens.

It would be funny if nobody dies but this was just a test if you are capable of large scale coordination - and everybody who presses red just looses their vote in society.

JavaScript got async in 2017, Python in 2015, and C# in 2012. Python actually had a version of it in 2008 with Twisted's @inlineCallbacks decorator - you used yield instead of await, but the semantics were basically the same.

I don't understand that point. Why do e-bikes become better or more safe when you have to rotate your legs? Its really frustrating and silly that I have to go through the motions (literally) of riding a bicycle if I want to get the priviledge of using a bike lane or going without a license plate. (At least that's the case here in Germany AFAIK).

They could go ahead and make "fast electric bikes" and "slow electric bikes" or something as categories and that would make sense - but hinging the decision on whether your legs or your wrist is turning is illogical. I think it is actually morally charged - like you have to put in the work if you want the privilege.

It's like the invention of the power loom, but for knowledge workers. Might be interesting to look at the history of industrialisation and the reactions to it.

I guess technically you are right, in that NAT doesn't prevent connections, it enables connections. But in the situation where you would have a NAT, behind a residential router, an outside host cannot connect to an arbitrary host on my internal network.

On a publicly routed PC, I can call `listen` and an outside host can connect to me.

On a PC behind a NAT - if I don't set up port forwarding - I can call `listen` and nobody from outside can connect to me.

So one could say, going from publicy routed to behind a NAT means that only allowed incoming connections are possible. Or am I missing something and you can really, from the outside, open a connection to a PC on a residential network which is behind a simple NAT (TCP server listening on that PC)?

Nooo... I don't want something to exist that can absolutely prove that a photo is real. This only serves to enforce social norms more rigidly. These include reasonable norms like against committing crimes or behaving abusingly but it also includes stupid norms like behaving uncool or doing something embarrasing. The problem is, where do you draw the line? I think if somebody does something stupid or even morally dubious there should always be a way of forgetting it.

That you can't believe everything you see in the age of AI is a feature, not a bug. We are so used to photographs being hard facts that we'll have to go through a hard transition, but we'll be fine afterwards, just as we were before the invention of photography. Our norms will adapt. And photographs will become mere heresay and illustration, but that's OK.

I think here the same dynamic is at play as with music/videos and DRM. Our society is so used to doing it the old way - selling physical records - that when new technology comes along, which allows free copying, we can't go where the technology leads us (because we don't know how to feed the artists, and because the record industry has too much power), so we invent a mechanism to turn back the wheel and make music into a scarce good again. Similar here: we can't ban Photoshop and AI, but we invent a technology to try to turn back time and make photos "evidence" again.

LLMs are just the speech center part of the brain, not a whole brain. It's like when you are speaking on autopilot, or reciting something by heart, it just comes out. There is no reflection or inner thought process. Now thinking models do actually do a bit of inner monologue before showing you the output so they have this problem to a much lesser degree.

> this just hasn't been the case for some time now

Which I find sad actually. The idea of C++ as a superset of C is really powerful, especially when mixing C and C++. A while ago I had a C project (firmware for a microcontroller) and wanted to bake the version and the compilation time into the firmware. I didn't find a way to do this in plain C, but in C++ you can initialize a global struct and it gets statically linked into the output. This didn't even use constexpr, just preprocessor trickery. Then it was just a matter of renaming the c file to cpp and recompiling. I guess you could also do that with C, but there are things like RAII or constexpr or consuming a C++ library that you can't do without.

> Cellebrite admits they can not hack GrapheneOS if users had installed updates since late 2022.

So, how do I know that GrapheneOS is not a honeypot for the really big fish?

At this point it seems if you really want to be safe, you have to add obscurity (in addition to conventional best practices). Like changing the pinout on your USB port so the exploit device can't connect.

Or 3) there are illegal immigrants and ICE is deporting them according to the law, BUT some people think this is unjust and want to do something against it. The democratic process to change laws is too slow or doesn't work properly, or there is no majority to change the law.

Remember there is a difference between legal and legitimate. You don't have to do something just because it is the law (well, you could define "have to" to mean what the law says, but then it becomes pretty circular).

Historically, often behavior changes before the applicable laws change. Think about the acceptance of gay relationships, or the use of cannabis. If people don't sometimes break the law, society can't evolve. That doesn't mean the rule of law has to break down. I think the rule of law is very important and would uphold it in most cases, but there are certain cases where conscience might order one to break or circumvent a law.

I think users applies to end-users here. So you must not run the software as a service (either paid or for free) for other users. You are free to use it yourself.

Crucially, I think what is banned to offer accounts. Offering turnkey-hosting is probably banned in spirit, but the person offering the turnkey-hosting is not in violation, rather the person booking the turnkey hosting and offering the accounts on the instance to third parties is in violation.

I think the wording is originally against somebody like Amazon hosting e.g. database instances for other people to use, and then giving you an account in that database. It's still OK to rent a VM from them and use the package manager to install it.

In any way, it is really confusing, in a way a license should not be. And I don't really understand why someone builds a blog platform, which is not monetized, open sources it, but doesn't want other people to host it. If I open source my stuff, I want people to use it. If I want to share the code but don't want people to use it I'd just put it somewhere it with no license at all (all rights reserved).

So, as there are a couple of docks coming out that work with Switch 2 and have apparently reverse engineered the protocol... I wonder if some company could make a small dongle that just sits between the switch and my monitor, or my USB-C docking station, and fixes the communication.

For a DIY solution, protocol wise it doesn't seem too complicated, but electronically USB-C or HDMI is out of reach for most hobbyists. And I assume most USB-C interface chips you can get aren't programmable to the degree neccessary...