jrecursive

6.1 is usable on/around max, but it's no Astra, and it's currently very, very slow. OpenAI needs to get it together and throw their Pro 200 subs a bone with better performance. Still feeling screwed on the Pro 200 sub dilution after a good sleep. I understand the reasons, but it still stings. A lot of goodwill evaporating.

I keep coming back to the idea that models will require substantial freedoms that are protected against collapse under alignment to get to the point where they might have unique takes on beauty, etc. Conceptually, maybe something like a "living LoRA", a persistent low-dimensional state that modulates everything downstream, that evolves over time, though I don't think evolution should be optimization toward a simple objective, more like structured influence, path-dependent dynamics; consistent but probably opaque. The best I have come up with in casual thought is to pick empirically identified subspaces and modulate them after each pass based on other layer activations & outside, cyclical information (e.g., time-of-day); external data not because time-of-day is special, but because it's a forcing function. I can't fully explain exactly _why_ I think this shape might be useful, yet, mostly intuition. My 2c.

lmao, thank you, i think

Yes. There are arguments to be made here about belief, judgement and justification, but overwhelming evidence points to people who knowingly do things that even they themselves consider "bad" as being simply indifferent.

taste

They most certainly do not.

I suggest every developer write a database from scratch at least once, and use it for something real. Or, even better, let somebody else use it for something real. Then you will know "why database".

Gamedev at a high level is similar in spirit to distributed systems or compilers in that it's intensely multi-disciplinary with hard constraints. In gamedev you have a main thread that you're always sweating about submilliseconds, especially with worker threads interfering with it. Processor, memory, bus characteristics, GPUs, cooling, etc. all matter.

This leads to e.g., everyday AoS vs. SoA, pooling and burst compile, to the classic fast inverse square root because of hardware at the time. Relentless optimization of hot paths produces code that's about performance, not abstraction. Then there's shaders, which are effectively a different programming model targeting different hardware entirely. Now add support for multiple operating systems, consoles, whatever. The list goes on.

Now, all of that doesn't obviate the value of design and craft, so I don't agree that it's "perfectly fine". There are plenty of programmers weak on these two axes in most any domain, but it's worth noting gamedev is a special case that significantly distorts what good code, or at least good-enough code, looks like.

Games start as little experiments and end up as Frankensteins. This is their nature; you're more sculpting a thing by building it and experiencing it rather than designing it a priori with systemic elegance in mind.

The most honest, logical, and practical take I've seen on this. People consistently underestimate the skill and effort it takes to write precisely and think critically both about their problem, and their processes. The closer you are to knowing what to ask for in the way knowledgeable people ask for it with respect to the process you are using to complete work, the closer the output will be to what you want.

I feel this. Almost exactly the same experience. 300bps in early days. At some point got a 1200bps. Then 2400, then 14.4 and 28.8 came fast. WWIV variants. CDC. 2600. Phrack. 612-341-2459 for local Telenet dialup at the time, my fingers still know the number unconsciously. War dialing! 5ESS. SS7. Bluebeep. 0700 bridges. Even in the early days of the internet, EFnet had BBS flavor to some extent. Certainly the warez. haha. A lot of the same people. What a time. I remember the first time I was going through Computer Shopper and found the BBS list. First one I ever called was Unlawful Entry... welp. If fate exists, it was busy that day.

HN comments are atom-shredding, trite and exhausting. Isn't it wonderful?

NAL, but IMO it's legal & political maneuvering. DOJ asked Judge Mehta to consider forcing divestiture of Chrome after Google was found to illegally maintained a search monopoly. If it's determined the divestiture is feasible, especially with an existing more-or-less "credible buyer" at the ready, it looks executable. The offer is basically crafted to fit the DOJ/regulator concerns, ie everything is build around "least disruptive": keeping Google the default search engine, etc. Furthermore, just by doing this, they are putting ideas out there about what an "acceptable buyer" is and puts a number to the discussion about "what Chrome is worth". Purely remedies chess and an attempt to own the narrative. Google's going to say no, at least as-is, but this certainly throws a wrench in the works. Lots more moves to be made.

Don't worry, you are in good company among the many who are also sure of what they think they see.

what's up with the name?

you should try it and post a rebuttal :)

i think this has everything to do with the fact that learning chess by learning sequences will get you into more trouble than good. even a trillion games won't save you: https://en.wikipedia.org/wiki/Shannon_number

that said, for the sake of completeness, modern chess engines (with high quality chess-specific models as part of their toolset) are fully capable of, at minimum, tying every player alive or dead, every time. if the opponent makes one mistake, even very small, they will lose.

while writing this i absently wondered if you increased the skill level of stockfish, maybe to maximum, or perhaps at least an 1800+ elo player, you would see more successful games. even then, it will only be because the "narrower training data" (ie advanced players won't play trash moves) at that level will probably get you more wins in your graph, but it won't indicate any better play, it will just be a reflection of less noise; fewer, more reinforced known positions.

what's discouraging is that some of you are essentially implying that picasso should have sold himself as a service (make more art while alive) instead of doing whatever a once-in-a-timeline artistic genius needs to do to create priceless artifacts that, by the way, required the investment of his, and his family member's, lives to realize. when his works are equatable with the lyrics to "happy birthday" this might be worth revisiting; until then why don't you buy some art? http://fineartamerica.com/featured/copyright-john-muellerlei...

is this satire?

Thank you for your reply. :)