I wonder if having at least one eye on the ball might be their secret sauce.

I have to think right now there are people at nVidia (and also AMD) saying “gamers are a buzzing insect like distraction” when every barely-consumer-afforbable $750 “midrange” GPU they sell represents some fab capacity that could have been part of a five-digit AI component.

The optical turning point is here for a lot of things.

The non-US world is seriously considering Chinese cars, which was always a high burden to clear (see Japan and Korea getting their manufacturing together enough to field a competitive product)

I suspect we’ll see some flailed attempt to block Chinese RAM from the US market (cf. The router fiasco) but this one might have enough corporate inertia to sail through. When every business in the country can’t afford even basic 16GB office desktops, Intel, Dell, HP, and Apple are probably going to be making compelling arguments to their captive legislators. This might be the only way to avoid huge losses on their non-datacentre product lines.

I prefer the emotional comfort of a toroidial power transformer.

I do want to see where China goes with soft power.

I’m ashamed about the amount of my life I’ve wasted in Genshin and Where Winds Meet, but it still feels like the a lot of their cultural exports haven’t cracked the “just exotic enough for broad global appeal” that big-brand anime or K-pop have hit.

If nothing else, imagine they treat it with the same strategic importance they gave industrial dominance. There would be a room full of very stern and sincere Vice-Ministers for Waifu Development and Fanboy Accelerstion.

Use a shared external source of randomness, like one of the shortwave “numbers stations”, to seed an agreed upon algorithm that simulates shuffling a deck and dispensing the cards. You’re running the same code on the same data, so you could show the cards you had seen after each hand to confirm everyone is in sync.

Wouldn’t two half-sized sticks perform better?

I thought dual-channel RAM was a thing at least as far back as the original Socket A nForce chipset.

Yeah, but it was the norm prior to 1909 and was a little more aspirational than “he just got perforated, let’s slap his face on the half dollar.”

What’s sad is that the obverse of the 250th dime is probably the best of the redesigns. For the first time since 1946, we’d have a circulating coin with an idealized figure of Goddess Liberty instead of a dead politician.

The typography is atrocious though. And the lack of a cohesive design language sucks. The Charles III coinage from the UK are so much better as a thematic set.

Then maybe the West shouldn’t have spent 50 years selling the PRC the rope they’re using to hang them with.

Even some of the near east is starting to think about what happens when the goo runs out. Only the US is doubking down.

One huge issue is that LLMs do weird and stupid things differently than how humans do them.

If you’ve developed an eye for reading human-made changes, you’re not necessarily going to recognize new and surprising failure modes as easily. It’s literally harder than regular code review.

Humans with modern tooling, for example, rarely hallucinate field/class/method/object names because non-spicy autocomplete keeps them on the rails. LLMs seem much more willing to decide the menu bar is .menuBar and not .topMenu, probably because their training corpus is full of the former.

Many things still fall back to VGA, like old projectors.

Trying to sell consumers on “scaling solves everything” is going to be a hard sell.

If we look at general purpose computation, which had decades of actual scaling-solves-everything growth, you had two influences that made the message resonate with customers:

• Clear existing applications where more power made the experience straightforward better. Your spreadsheet took an hour to recalculate at 8MHz and 20 minutes at 25MHz. A lot of the “bigger model” stuff is plateauing with marginal or spotty gains. If I feed another 5 Internets of data to ChatGPT, will that summarized email be that much better?

• New applications that could be demoed on specialised low capacity hardware and scaled down to consumers as more power became available. Think of early CGI on hardware costing tens of millions, and now you can run Blender on a $149 laptop. Since most commercial AI plays are hosted services, there’s not much opportunity to tease that way anymore.

I figured SDF were the sort of people who would consider uptime a badge of honour.

I should probably be on Hexbear but I didn’t know of its existence at the time. Very problematic, my Xibuck direct deposits keep bouncing.

Transhumanity would be exciting if they had cool visions. I’d be all over raising a creche of draconic children.

But no, it’s just rich people gluing a Palm Pilot to their cerebrllum or doing a dance to shoo away the reaper.

The difference was that Amazon knew how to make a profit, but was reinvesting into infrastructure plays and bigger fish.

If they had to, they could have been a modestly profitable bookshop in 2002. AWS and monster logistics might not have developed to put them in the 13-digit club though.

Does any AI-centric play have that fundamental fallback? The services that seem to be most effective at direct monetization, the coding tools, are typically running at huge losses. If they raised costs to cover, precious few firms will pay basically the salary of a senior dev for an emulation of an enthusiastic junior dev with an affinity for footguns.

The less enterprise-focused products-- parasocial toys, image and video gen, will likely try to dip into consumer subs and advertising, but can that generate the cash volumes these platforms demand?

All the best firms continvoucly morg these days.

The “default” mode for a USB keyboard allows submitting 6 keys + modifiers. Some boards define nontraditional input descriptors that allow more, but that mode is not guaranteed to work in places like the BIOS menu or naive KVM switches.

To avoid phantom keypresses when you hit three keys in a “square” on the matrix, a diode can be placed in series with each switch so current can’t go through an “indirect” route.

There’s also an execution problem.

Truly knowing your customer might produce very different outcomes than the current compliance checkbox approach.

“I know Fred just sold his old car. The idea he suddenly has $12k in cash is not suspicious” or “Jane’s been talking about going to Montreal for momths. We should not block her card when it lights up there.”. That’s real KYC, but it requires human connection and human judgement, which doesn’t scale and doesn’t provide the right paperwork for demonstrating compliance with arbitrary mandates.

I think she did a poor job of saying what she brought to the table. I understand not wanting to throwJoe under the bus, but opening some daylight on policy would have given her a chance to deflect the affordability problems the last months of Biden had, for example.

Culver’s is… interesting.

I find I don’t like the paper-thin style burgers much (though the fish sandwich is probably the best fast-food fish sandwich on the market) but it always gives the impression they’re trying a little harder. The restaurant always seems clean and a bit of effort was done on the appearance, like they’re still in 1978 and taking the family for a sit-down meal there might be an option.

OTOH, the customer base seems to be people who have been going there since 1978, but that could be my location.

They’re opening a new one 3km from my house, next to a McDonalds that hasn’t made a single order correctly since it opened in ~2015, so it will be interesting to see how the market shakes out.