A lack of tool & die isn't the West "forgetting how to make things" so much as it is the West choosing not to make the things that make things.
To the extent that AI is analogous to automation in manufacturing, and "writing code" to working on an assembly line, it's hard to argue the West is any thing other than a global leader in "software tool & die", so to speak.
Been meaning to check out Railway for a while, but now feeling happy about dragging my feet.
As flashy as their DX seems to be, the fact that a sketchy single VPS node with a server, a SQLite instance, and a LiteStream hookup has a better recovery story really makes me not trust their platform.
What sells me on it is I get to take a spare Gen4 m.2 ssd out of my gaming PC I wasn't fully utilizing instead of paying for 1TB of storage.
Being able to drive the price down by re-using parts I already have is a pretty big selling point, IMO.
Also, I think Apple is benefitting from scale, since they're able to maintain the (usually too high) storage and memory prices they've had for years. At this moment in time, framework have the misfortune of being forced to pass inflated wholesale prices onto the consumer.
Make this comparison one calendar year ago and the F13 Pro could very easily beat the MBP on price spec-for-spec.
This seems huge for subscription customers. Looking at the Artificial Analysis numbers, 5.5 at medium effort yields roughly the intelligence as 5.4 (xhigh) while using less than a fifth the tokens.
As long as tokens count roughly equally towards subscription plan usage between 5.5 & 5.4, you can look at this as effectively a 5x increase in usage limits.
As someone who always leaves intelligence at default, and am ok with existing models, should I be shifting gears more manually as providers sell us newer models? Is medium or lower better than free/cheaper models?
For some real data, Artificial Analysis reported that 4.6 (max) and 4.7 (max) used 160M tokens and 100M tokens to complete their benchmark suite, respectively:
Looking at their cost breakdown, while input cost rose by $800, output cost dropped by $1400. Granted whether output offsets input will be very use-case dependent, and I imagine the delta is a lot closer at lower effort levels.
If your use-cases don't benefit from RSC performance characteristics then they probably aren't outright better.
But I do think they're a compelling primitive from a DX standpoint, since they offer more granularity in specifying the server/client boundary. The TanStack Composite/slots API is the real selling point, IMO, and as far as I can tell this API is largely (entirely?) thanks to RSCs.
Right, but the picture those statements painted collectively was not flattering. And that was certainly intended by the authors. Thus, critical, but not at all "incendiary."
Update: To clarify, my personal stance is that the critical tone was both intended by the authors and, in my opinion, appropriate given how much power Mr. Altman holds. If he has a history of behaving inconsistently, that deserves daylight.
Are you arguing that because the authors knew the pattern they were documenting was unflattering, the piece is somehow compromised? That they clearly had an agenda? That's called reporting. They called a hundred-plus named sources and the picture those sources independently painted was damning. Altman has a history of telling repeated, easily-checked lies, followed by fresh lies when caught in the first ones.
Are you suggesting that they should have "both sides"-ed by reporting company PR and Sam-friendly sources and giving them equal weight? Sometimes the facts point in one direction.
> Are you arguing that because the authors knew the pattern they were documenting was unflattering, the piece is somehow compromised?
Uh, no? Lol, I'm on your side, bud. Put away the pitchfork. I thought it was a really good and fair article. I am not the adversary you're looking for.
> my personal stance is that the critical tone was both intended by the authors
You may think we are on the same side. You don't understand what side I'm on. "Lol".
Your "personal stance" is that you can get inside the heads of the reporters? Obviously not. So you're going by the idea that an article that leads to critical conclusions is inherently slanted. This is an insidious and damaging idea. It has led to the belief by journalists and editors that they need to twist themselves into pretzels to present "both sides", which is easily exploited by people of bad faith to launder outright lies. There's a direct line between this and authoritarianism. I'm quite serious about this. The fact that you agree with the authors in this case is completely orthogonal.
Every article is inherently biased due to the fact that there are inclusions and omissions. This is just a fact.
You're injecting your own personal view into GP's statement by adding a lot of weight into the distinction between the words "critical" and "incendiary" and "neutral", when GP made a very neutral and not as charged statement.
I love reading stuff like “Critical, slanted, and compromised mean the same thing. They are interchangeable words.”
Given that, it looks like your position on davesque’s posts is slanted. Your take is critical of those posts, which means your assessment is compromised, and as such should not be taken as valid.
And I love seeing sentiments attributed to me, in quotes even, that I didn't state or imply, and certainly don't believe. "Critical" by itself is not a synonym for "slanted". However the post I was commenting on was:
> Right, but the picture those statements painted collectively was not flattering. And that was certainly intended by the authors. Thus, critical, but not at all "incendiary."
The key there is "certainly intended by the authors". The full sentiment here IS equivalent to "slanted".
My (potentially naive) take is that open models will save us. The biggest markets for LLMs (e.g. coding) are narrow-enough to be served well by smaller models with proper RL. Cursor's Composer 2 (created from a Kimi K2.5 base) is a great example, and I expect it to be the first of many.
The wealth of great open models provide an excellent base for fine-tuning, distillation, and RL. I see a lot of untapped potential in the field of bespoke, purpose-built models that can be served far more cheaply than the frontier competition. I would not be surprised if we see frontier-adjacent experiences running comfortably on a Mac Mini by year end.
With frontier models seemingly hitting diminishing returns in quality, I struggle to see a world in which gigantic, expensive, general-purpose models don't become increasingly niche.
To the extent that AI is analogous to automation in manufacturing, and "writing code" to working on an assembly line, it's hard to argue the West is any thing other than a global leader in "software tool & die", so to speak.
reply