I would not bet against it, that’s for sure.

My bet is on j before Starfield 2. BGS will really, actually think “hey, let’s remind everyone how great Starfield 1 was!”

I don’t think it was a sales failure, was it? Not a short term one.

At risk of going Fallout 4…

How about medieval homebuilding?

Build a little hut? Store stuff. Have chickens. Fish in a pond out back, kiddo running around.

If it’s done even half as meticulously as KCDII, I personally know people that would play the hell out of content like that. That’s the premise of a whole lot of Skyrim modding.

Yeah.

Can you imagine a “Starfield: Special Edition” re-release in a few years?

I’d say the notion is satire… but their leadership’s ego is so titanic it might not be.

I dunno. I saw a dev video where the project leads, Henry’s VA, and other team members all seemed excited about the development expansion.

…Maybe they’ve just done the hard part already?

KCDI had issues, but they knocked KCDII out of the park. CryEngine is utterly fabulous for that niche; they don’t need to change a thing, mechanically or visually, nor worry about maintaining a whole game engine; that’s offloaded to Crytek.

Maybe they’ve really got all the tooling and workflow perfected for their “Very medieval RPG” template, and hence can make another banger without so much R&D and fuss.

One thing I like to bring up:

If current LLMs are so intelligent, can they write their own inference code?

…Or even make modest code contributions to those bases?

The answer is a resounding “no.” See: llama.cpp’s contribution policy, vllm’s PR history, janky diffusion UIs and such.

The severe bottleneck to efficient inference (much less training) of a variety of models is good code contributors, and AI is doing little to help. And, ironically, if it ever was smart enough to do that, ChatGPT the service would collapse in like a week because a ton of backlogged integration would get done, and the open source/open weights space would explode.

It was just about perfect. Andromeda’s MP wasn’t even bad, but somehow not in that sweet spot.

ME3MP is still active, too. You can find public lobbies, and there are organized events.

Nostalgia is huge with them.

I loved Oblivion. I used to appreciate players being into the lore, the little corners of the game. But then Starfield came out with so many defenders. It made me realize the sheer number of players that are just… I dunno. “Shills” doesn’t feel like the right word, but I do not understand why they go to such lengths to worship BGS.

They’re all lonely and, yes, a little grindy to me without multiplayer to “show it to someone.”

That being said, I may have been burnt out by modded Minecraft when I was younger.

Right?

I played the hell out of ME3MP. I played some Warframe. I get the “PVE shooter zen”

But WTF. Everything I see and read about Destiny 2 sounds infuriating, and I can’t comprehend why people put up with it.

I disagree here. The side quests and atmosphere are great. FemV nailed the voice acting. It has a lot of fun, bitingly memorable characters I still think of, sometumes.

The gameplay is satisfying… if, and only if, you mod it to not be bullet spongey and so grindy.

I’m not a 2077 apologist.

Does it run well? Nope.

Was it absolutely unacceptable at launch, and years after? An understatement.

But (in 2025/2026), it deserves the praise it gets, IMO. With a few mods, it’s a good game now, not just “a good game somewhere in there.”

Toms has always been clickbait. They just copy and sensationalize other headlines they find, and they’ve frequently bent the knee for Nvidia. Other outlets used to make fun of them all the time.

It’s sad they “survived” the enshittification of the internet and people keep sharing their clickbait :(

That being said, there’s some truth here.

MTT is just a pipe dream, last I checked. But Deepseek is actively being served, in mixed FP8/FP4, on racks of Huawei accelerators.

I believe Baidu trained a model on them, too. But most training (like Deepseek’s) is still done on CUDA.

…Also, be careful equating this stuff with any kind of “consumer friendly” hardware you or I could buy. That’s less likely. The Huawei accelerators (and other local Chinese hardware experiments) are geared towards huge servers serving requests in parallel.

Perhaps ideological purism was never a good idea?

…Like history suggests?

Hence my radical preference: “a la carte” economic systems. Shameless capitalism, frothing communism, anarchism, authoritarian technocracy, even theocratic systems, they all excel in certain sectors and not others. Sometimes, in nutty combinations.

So why idealize one?

A “mix” has been reality for a long time, anyway, but I think that should be embraced more explicitly. Get systems where they’re good instead of shoehorning them where they’re horrific.

YMMV.

You gotta look up reviews. Both buyer track records, and focused reviews that look at what controller/NAND they use.

Many SSDs are basically the same as other brands internally. A few are wonky. It just depends.

As an observer in these comments, this is a great answer. Thanks for typing it out.

It does seem like some “cuts” could be ironed out reasonably quickly, like the file naming issue or UI lag.

The Chinese AI labs are really trying to pop the bubble, too.

How?

Well lemme ask you this. What if models 80-90% as good as Claude, with weights just thrown out there for any provider (or homelab) to host, flood the market? What if they’re so dirt cheap to run, they’re almost free, and don’t even need Nvidia GPUs? What they need fewer resources to run with each update, instead of more?

…What if this already happened, and Big Tech is maddly lobbying to ban/censor them before people realize it, and that the “infinite scaling” thing is a big fat lie?

That’s the state of things.

Even the machine learning dev/tinkerer communities hate AI.

They were making fun of Sam Altman (or calling him a sociopath) before it was cool.

Somewhat related:

” You need to understand that Sam can never be trusted. He is a sociopath. He would do anything.”

• Aaron Swartz, just before he died in 2013.

That’s only partially true.

The other half is that “almost as good” AI like GLM’s coder plan, Kimi, and dozens of other providers are dirt cheap. And getting cheaper, fast.