From "Jokester" by Isaac Asimov 1956:
"Early in the history of Multivac, it had become apparent that there was one big bottleneck: the questioning procedure. Multivac could answer the problems of humanity, all the problems, if -- if it were asked meaningful questions. But as knowledge accumulated at an ever-faster rate, it became ever more difficult to locate those meaningful questions."
[0] https://web.archive.org/web/20150118004835/http://www.sffaud...
I'd say that a more appropriate reference from that time would be "The Nine Billion Names of God" by Arthur C. Clarke [1], which actually deals with the finiteness of the list of problems that a machine successfully exhausts.
[1]: https://hex.ooo/library/nine_billion_names_of_god.html
>>which actually deals with the finiteness of the list of problems that a machine successfully exhausts.
This is already the case with most DevOps jobs with Claude Code. There are only that many finite issues that need fixing in production, even for very large systems.
So you don't need all that many people to run ops these days.
Yet as a customer I don't think I've got to endure as many outages in my life before 2026 as I did for the past 9 months…
And it's not just github, everything seems to be down on a daily basis right now.
Seriously, it feels like services that were once rock solid are now trembling every time the wind blows.
If we're being anecdotal, I haven't had a single outage of anything affect me at all this year, but have had many in previous years. Often the thing blamed publicly has been incorrect too.
I am kind of saddened by technology in general that we need LLMs to deal with it in the first place.
It used to be that you compiled an executable, copied it onto the target computer and it worked. Then came servers, where you had to open some ports, set up some init scripts, maybe a db connection string.
This has evolved into a fractal of awfulness of a myriad of options and problems, each of which more elaborate and specifically limited than what came before, so you can't really use any one of them, but a combination, which is bound to interact weirdly, giving you some unique problems.
LLMs are great at solving these kinds of problems, but these problems should not exist in the first place, and they just waste everyone's time.
When it comes to actual meaningful product questions (and not just 'how/where does this code run'), LLMs are basically next to hopeless.
Since LLMs are very good with dealing with the BS, I hope they'll be useful in making people recognize and do away with BS altogether, and we can go back to the simple olden days.
Software's always been painful to run and use.
>It used to be that you compiled an executable, copied it onto the target computer and it worked.
Java had been marketed last century as a 'write once, run anywhere' language. Tells you a lot about how portable executables in other languages were.
Just branch to the first instruction, and let it take it from there. The shackles exist but in your mind.
Well you see we need to migrate system A to B and X to Y and here's another new integration to look at and we need to finally get off of this EOL distro and... It'll never stop.
I don't see why that should be a problem, as we already know the answer is 42 in any case.
There is as yet insufficient Data for a meaningful answer.
[1]: https://www.imdb.com/title/tt0708807
??? http://www.thelastquestion.net/
The findings of the present study suggest that more funding is necessary, roughly the amount of money necessary for a trip to Burning Man
You may recall though that whole problem with '42' as the answer was the lack of knowledge of the question. The Earth was created as an attempt to provide the question, but was unfortunately demolished to make way for a hyperspace bypass just before the question was resolved.