Comments on: People who understand complex systems also understand the importance of minimising that complexity wherever possible

By: T. Beholder

T. Beholder — Sat, 18 Apr 2026 18:11:55 +0000

Jim says:

LLMs now provide (a) and (c)

It also provides (b) to some of the Pointy Haired Bosses, if as a mixed blessing.

In that the big downside is having much fewer subordinates involved in BS production, but the upside is perfect excuse for a purge and then having much fewer potential backstabbers.

By: Jim

Jim — Wed, 15 Apr 2026 02:39:24 +0000

T. Beholder: “For one, let’s remember that more powerful computers did not help to write and optimize a better code, but rather allowed greater inefficiency. Thus lowering standards for software to the levels that would be considered absurd but a few years earlier. The Daily WTF found IIRC embedded Java machines whose only purpose was to perform simple arithmetic operations with string constants.”

The objective functions of that code were and are: (a) something that looks like it works (and mostly does), (b) job security, and (c) rational apathy.

LLMs now provide (a) and (c) to goyim in abundance.

By: T. Beholder

T. Beholder — Sun, 12 Apr 2026 19:58:45 +0000

He is almost certainly wrong on LLM keeping the verbal lawns shapely.

For one, let’s remember that more powerful computers did not help to write and optimize a better code, but rather allowed greater inefficiency. Thus lowering standards for software to the levels that would be considered absurd but a few years earlier. The Daily WTF found IIRC embedded Java machines whose only purpose was to perform simple arithmetic operations with string constants.

So instead it’s one more way so-called AI ends up as a «force multiplier for stupidity».

LLM can navigate and parse a vast, ever-growing mess? Great! The pointy-haired bosses can and will use it this way. And continue not caring about growth of the contradictory mess. But GIGO principle, of course, still stands. And all data going in (semi-articulate requests and cancerous databank) is mostly garbage.

The inevitable result: LLM used as a black box oracular tool with pseudorandom input.

Lucklucky says:

Not if their jobs, power, and status depend on complexity.

If Parkinson’s Laws remain part of the context, some tool is not going to make them disappear in a puff of logic smoke, yes.

By: Lucklucky

Lucklucky — Tue, 07 Apr 2026 20:59:36 +0000

Not if their jobs, power, and status depend on complexity.

By: Gaikokumaniakku

Gaikokumaniakku — Sat, 04 Apr 2026 22:15:31 +0000

People who focus on receiving trust from authority have less attention available to learn engineering. People who focus on technical craft have less attention available to learn how to gain trust from authority. Thus the managers (in my experience) are always technically clueless.

Hackernews recently had a discussion comparing personal coding projects to the Winchester Mystery House.

https://news.ycombinator.com/item?id=47601194

This links primarily to :

https://www.dbreunig.com/2026/03/26/winchester-mystery-house.html

But crucially in that discussion, “mrandish” mentions Bill Atkinson’s “negative lines of code.”

https://www.folklore.org/Negative_2000_Lines_Of_Code.html

Bill Atkinson clearly worked to reduce complexity without reducing functionality.

Also in this discussion, some interesting projects are mentioned, such as “Gastown,” which reminds me of “Bartertown” from Mad Max.

https://steve-yegge.medium.com/welcome-to-gas-town-4f25ee16dd04