HN user
I'm an EE and SW dev with experience in ML, software, and hardware design.
Definitely sympathetic to their policy, but AI tooling and quality are changing quite fast. In a year I'd expect a modification of this as AI agents get better in virtually every possible way.
Anybody remember corewar, where each player wrote an adversarial programs in redcode?
Wow. Where was this in the 90s when I "needed" it.
Is there a console version?
Totally agree.
Agree.. 68k assembly was dreamy compared to 80x86..
Wow.... I remember writing 8086 assembly on MASM and another assembler I've forgotten the name of, and then also doing inline ASM in Turbo C++
The segment thing and the convoluted different pointer math caused real gymnastics if you ever had data bigger than 64k... such as images.
I always thought of the segments as windows of 64k but moving between those windows, esp with the limited register set, required some real mental gymnastics.
There is some base level of intelligence any model needs to be useful, even in narrow tasks.
Could you teach a 5 year old to drive a car? A 10 year old? A 12 year old? To drive a car requires being able to read, to have judgement about ice or rainy conditions, to anticipate a child running after a ball. By the time a human in in their mid teens they have acquired the base knowledge...
Small models need to have enough base knowledge to be able to be good enough -- even in a seemingly narrow regime. Where is that? Obviously they don't need all the obscure knowledge of a frontier model but there is some base level which is probably more than it would first seem.
As smaller models become more capable I suspect we will see a reset in token economics.
How long has this been running?
In word.. the asks need to separated from execution. Labeling or tagging the prompt itself is a dead end.
It seems useful but then we have to manage similar metadata in multiple places, so hygiene around consistency becomes important
Truly agree with the framing of buy vs build.
Also, some software businesses use a ton of aggregated or hard to get data which needs to be synthesized and that doesn't go away even if the llm driven coding is cheap.