3 ms·

So your claim is that anything turing-complete should forever be called "tech" and that any disagreement is arrogance? That's a difficult position to engage wi

So your claim is that anything turing-complete should forever be called "tech" and that any disagreement is arrogance? That's a difficult position to engage with constructively.

ggm 6y ago
At some future date, assuming the GPT-3 adherents are right, nobody will "write code" because we will describe systems design goals to an intermediary which then writes the code. Where is the tech then? I do think there is tech and non-tech. But it gets very like pornography and the Meese commission: I know it when I see it. What I find hard, is how many people writing for the REPL or compiled languages, dismiss what is done in CSS or in JS outside of webasm. or PHP. The arrogance is the exclusionary behaviour. Carmack is very smart. Had he been born 40 years earlier he might well have been the one to work out how to optimise track to head delay on a drum memory, or hacked delay lines, or even have done something amazing in LSI before VLSI. But then some non-carmack doing bitblt optimisations to make games happen would have been wondering "am I a real tech or not" (btw I think GPT-3 Adherents are smoking strong stuff. I don't expect anyone to write informally stated natural language of intent and have proscriptive, defined code come out the other side, but you know this is a strawman argument too) I graduated in 82. Because I did not do the Mudge VLSI Course I was held in somewhat contempt by my peers who did. I did Pascal. I was held somewhat in contempt by the Fortran NAG library maintenance squad. We who ran the boxes, and wrote controllers for the plotter were a bit contemptuous of the ones coding for the BBC micro and I was completely distainful of the guy coding .BAT files. None of this is "good" behaviour because we were all doing "tech" But the guys in the engineering labs doing Hypersonic wind tunnels were sometimes a bit elitist about us in s/w ... Over in the school for the logical foundations in computer science they were already writing abstract meta-code which was as far removed from an instruction set as its possible to be: theories of computation, proof systems, you name it. Physicists and Engineers have been arguing for centuries about the difference between their disciplines.
- AlchemistCamp 6y ago
  Ah this explains the disconnect. I see the "Web 2.0" era as closer to the frontier than Quake 3 (as cutting-edge as it was in its day). In the era of Flickr, a young group of devs with no resources would have had a tough go at it in 3D gaming and would have been better off making Flash games. I can't say much about if and when we're worse at everything than AI, but we're not there yet. At least for now, there are still new opportunities just as open to aspiring techies now as web apps were 15 years ago or PC games were 30 years ago.
- imtringued 6y ago
  >At some future date, assuming the GPT-3 adherents are right, nobody will "write code" because we will describe systems design goals to an intermediary which then writes the code. Well, the fundamental problem with this is that you just made the programmer even more powerful. A single programmer is now as productive as a thousand programmers. You still want to have a programmer because you want someone who understands what the system is actually doing right now and keeps it running. You also need someone who understands the limitations of the software generator. You can't just say "please create software that solves the halting problem". You also need someone who understands how to optimize software. You can easily ask for something that is possible but it could cost you millions of AWS credits. There will always be someone who has intimate knowledge of the tool he is using and can charge appropriately for it. The more powerful the tool the more attractive the employee. Compare an engineer drafting on paper vs one doing it in CAD.

So your claim is that anything turing-complete should forever be called "tech" and that any disagreement is arrogance? That's a difficult position to engage with constructively.

ggm 6y ago

At some future date, assuming the GPT-3 adherents are right, nobody will "write code" because we will describe systems design goals to an intermediary which then writes the code. Where is the tech then? I do think there is tech and non-tech. But it gets very like pornography and the Meese commission: I know it when I see it. What I find hard, is how many people writing for the REPL or compiled languages, dismiss what is done in CSS or in JS outside of webasm. or PHP. The arrogance is the exclusionary behaviour. Carmack is very smart. Had he been born 40 years earlier he might well have been the one to work out how to optimise track to head delay on a drum memory, or hacked delay lines, or even have done something amazing in LSI before VLSI. But then some non-carmack doing bitblt optimisations to make games happen would have been wondering "am I a real tech or not" (btw I think GPT-3 Adherents are smoking strong stuff. I don't expect anyone to write informally stated natural language of intent and have proscriptive, defined code come out the other side, but you know this is a strawman argument too) I graduated in 82. Because I did not do the Mudge VLSI Course I was held in somewhat contempt by my peers who did. I did Pascal. I was held somewhat in contempt by the Fortran NAG library maintenance squad. We who ran the boxes, and wrote controllers for the plotter were a bit contemptuous of the ones coding for the BBC micro and I was completely distainful of the guy coding .BAT files. None of this is "good" behaviour because we were all doing "tech" But the guys in the engineering labs doing Hypersonic wind tunnels were sometimes a bit elitist about us in s/w ... Over in the school for the logical foundations in computer science they were already writing abstract meta-code which was as far removed from an instruction set as its possible to be: theories of computation, proof systems, you name it. Physicists and Engineers have been arguing for centuries about the difference between their disciplines.

AlchemistCamp 6y ago
Ah this explains the disconnect. I see the "Web 2.0" era as closer to the frontier than Quake 3 (as cutting-edge as it was in its day). In the era of Flickr, a young group of devs with no resources would have had a tough go at it in 3D gaming and would have been better off making Flash games. I can't say much about if and when we're worse at everything than AI, but we're not there yet. At least for now, there are still new opportunities just as open to aspiring techies now as web apps were 15 years ago or PC games were 30 years ago.
imtringued 6y ago
>At some future date, assuming the GPT-3 adherents are right, nobody will "write code" because we will describe systems design goals to an intermediary which then writes the code. Well, the fundamental problem with this is that you just made the programmer even more powerful. A single programmer is now as productive as a thousand programmers. You still want to have a programmer because you want someone who understands what the system is actually doing right now and keeps it running. You also need someone who understands the limitations of the software generator. You can't just say "please create software that solves the halting problem". You also need someone who understands how to optimize software. You can easily ask for something that is possible but it could cost you millions of AWS credits. There will always be someone who has intimate knowledge of the tool he is using and can charge appropriately for it. The more powerful the tool the more attractive the employee. Compare an engineer drafting on paper vs one doing it in CAD.