duskwuff

Re. 5:

LLMs do not generate images. LLMs prompt distinct, separately trained image models to generate images. The LLM has no ability to introspect the image model and cannot provide feedback during the image generation process. If the image model misinterprets the LLM's prompt (which can happen!) or inserts unexpected content, the LLM may become "aware" of that during subsequent chat steps as it ingests the generated image, but it cannot provide detailed control over their generation.

Quiz: What does the second "L" in "LLM" stand for?

Hint: it isn't "image".

> Why isn't the publishing of the captured video banned?

It could be, but that's a separate problem. Surreptitiously filming someone can be problematic even if the videos are kept private.

Traction control was allowed before 2008. The problem was that it trivialized some aspects of driving - drivers could floor the throttle through turns at little to no risk. (I suspect there may have also been some concerns that drivers who pushed traction control too far could have easily ended up in unrecoverable situations.)

There's some interesting discussion of the history and technical aspects at: https://www.f1-forecast.com/pdf/F1-Files/Honda/F1-SP2_25e.pd...

Unfortunately, I have a feeling that their "solution" would be every bit as weird as what McMansion Hell is highlighting. See e.g. https://www.dyson.com/air-treatment/fans-heaters/cool-cf1/wh....

Something, something, Torment Nexus...

Don't worry, they're still making it worse:

https://www.reddit.com/r/modnews/comments/1wubgvt/continuing...

(TL;DR: they're killing old.reddit, RSS, automod, and a lot of other bots.)

Components like transistors usually have to go on the surface of a PCB. (While there are techniques for embedding components in a PCB, they're extraordinarily expensive and would not be feasible for a project like this one.)

A replica 386 would be impractically large - it's got around 275k transistors, compared to the 3.5k in the 6502. At the same scale as the MOnSter 6502, that'd be about a hundred square feet of PCB.

The number of layers isn't even remotely close to being a constraint. The real i386DX was manufactured on a process which only supported two metal layers, and a PCB version isn't subject to the same constraints as silicon.

> I suspect that [modifying data in memory] is how the Game Genie for the Nintendo Entertainment System worked.

Game Genie, Action Replay, and other similar devices worked on a different principle - the "codes" represented patches to the code in the game cartridge, not to data in memory. (The original Game Genie for the NES could patch up to three bytes.)

In practice, these patches often amounted to simple modifications like changing the constant value used to initialize a "lives" or "health" variable, or patching out code which caused the player to take damage or die.

Output lane switching is possible - some mods (e.g. [1]) add it, as well as even wilder options. Not offering it in the base game is a deliberate choice.

As far as belt lane management goes, another interesting option - which is also present in the engine, but not used by default! - is lane splitters [2].

[1]: https://mods.factorio.com/mod/quick-adjustable-inserters

[2]: https://mods.factorio.com/mod/lane-balancers

For another example: one "theme" listed is "Walking Dead episodes". The streets which come up are ones with names like "Faith" or "Family", which happen to also be names of episodes of that show. There is clearly no actual connection here.

IIRC, a number of the team members started as players.

Those are aftermarket products for use in heavy vehicles, not something that's present (or usable) in passenger cars.

> Aside from memory safety, which is solved by using a compiler that just doesn't allow unsafe memory operations

I don't see how that's possible without turning the language into something that isn't C, either by adding significant new functionality (e.g. fat pointers) or subtracting enough functionality that it's a much less capable language (e.g. disallowing dynamic memory allocation).

Unfortunately, they can't:

https://en.wikipedia.org/wiki/Wolf_Amendment

> There are still some types of CRT-related vacuum tubes in production for applications digital equivalents are objectively less performant in.

Like what? The only significant applications I'm aware of for vacuum tubes nowadays are high-power RF amplifiers (including household microwave magnetrons) and intentional X-ray emitters.

> I was wondering how you could possibly sync up something like a Carrier that has multiple fighters.

Same way as everything else - you play out the simulation and see what happens. The behavior of the interceptors is as deterministic as any other unit. (They just happen to not be directly commandable by their owner.)

Replay files are streams of the orders given by players. They generally can't do anything which would be impossible in game.

Not feasible. The VODs are from a spectator's perspective; there's a lot of action happening "off camera" (like players building units) which affects what's seen on camera later, and which would be extremely difficult to reconstruct from what's visible.

What you're describing is a casual game, not competitive play. And, of course, all bets are off if you're playing with additional rules like "no rush" which remove the element of tempo from the game.

Yes, e.g. https://github.com/gulshngill/bwrepanalysis. But I doubt an agent would learn much from them without a lot of additional processing - the replay files are little more than a stream of the orders given during the match (e.g. "at tick 17, player 1 ordered unit 234 to attack-move to 56,78"). They're difficult to make sense of without a lot of additional context, like the map layout, the location and status of other units, what parts of all that are actually visible to each player, etc.

This is not a viable strategy in competitive play. It's a huge resource/time investment with a dubious payoff.

If you could win as Protoss by building Dark Archons, raiding your Zerg opponent's base to capture a worker, building a bunch of Zerg production and tech buildings, and attacking with a combined army... you most likely outclass your opponent, and could have won much faster using a more conventional strategy.

> They've gone with a lesser known company Kioxia.

Formerly known as Toshiba's memory division. They spun off the business as Kioxia in 2017.

> So why not set up a simple grid (say, Pages on a mac or Libre Office or Scribus) and pair the best photo you can grab with a description opposite it, on a nice background?

Or, for that matter - skip the photos, just do text, maybe with a few bits of clip art if you're feeling daring. Photos are hardly essential for coffee drinks, which tend to all look the same in real life anyway.

> It's also nice that it translates directly to the file system.

On the other hand, this means that it's bound by filesystem naming rules - e.g. you can't have an object named "abc/", or a pair of objects named "a" and "a/b".

> The problem with ebooks is that you cannot be 100% certain that you're reading the original text.

That's equally true of print books. Misprints happen.

1992. (It's a music joke.)

https://en.wikipedia.org/wiki/Automatic_for_the_People

Telegram's usage of "GIF" (any short video without sound) is incredibly weird and frequently goes against users' expectations. I wouldn't treat that as definitive.

Yes. These drawings are a companion to Apple's accessory design guidelines, which include a lot of other interesting design considerations:

https://developer.apple.com/accessories/Accessory-Design-Gui...