Hacker News Reader

Comment by ncmncm | Hacker News Reader

This is an astounding record of an astounding conversation.

So much died with them. We use control systems everywhere, but we should be seeing them everywhere, including the accidental ones, and fixing the ones that don't produce good results. Instead, control systems are boxed and understood only by specialists who only think of the box.

View on HN

I think we need to revisit "Software is eating the world."

There's a certain antipathy that's inherent to the current formula and it shows up in your 'specialists who only think of the box' comment.

In a way it relates to work-life balance. We should be able to have developers in their 30's who are specialists in several unrelated fields. Dive back into something you did before you went into the CS program, or something you found along the way.

Instead we blow right past our 10,000 hours toward 20,000 and a lot of us never take a detour. There are diminishing returns here. Other skills are also important (even more effective) once you know basically how to do your primary duties without having to stop and think about it.

To really eat the world, we need to participate in the world. 'Disruptors' usually think everyone else's problems are simple, due to pig ignorance. I'm not sure when the backlash is coming but I'm sure it's out there.

I watched a 'household name' in one of my hobbies do a kickstarter to create a public website with detailed information for that hobby. He got what sounded like a bunch of money but ended up having to make most of it subscription based. Guess what? Programmers are really fucking expensive. Especially if they work on the wrong things.

If I can get my ducks in a row, my side hustle will basically be replacing his bad execution of a good idea with one that works. I can charge myself 'wholesale' prices for labor instead of retail. I need fewer user studies. Hell I know what a user study is. And I can care better about things like privacy and scalability and reliability. For me it won't be a dancing bear. I can bring in some of my expectations from my day job.

My two biggest regrets in my education are (1) taking the 'college prep' classes in high-school instead of the welding and small engine repair style classes and (2) majoring in CS instead of minoring while studying a field like biology or chemistry that's full of things that need programs written about them (and looking down on the kids who did, because I was the superior math/CS student).

I mean, I don't think I have a bad life, but it's kind of a regret that when I sit down and think about things to program for fun, I think about theorem provers and compilers, which there are already bunches of.

Electrical Engineering, its very cool to be able to calculate and invent anything.

In junior high I took wood shop. I used a router to make a birthday plaque for my dad. He didn’t smoke so the stereotypical ashtray wouldn’t do him any good. I also learned how to work with acrylics.

In high school I took an auto tech class from my local community college where I rebuilt an automatic transmission. It still amazes me those devices work at all.

In college I took a welding class from the agriculture school for elective credit. Everyone should have the experience of cutting metal with fire.

Those experiences have been priceless to me in my day job at $megacorp. I work with a lot of people who have only had one job. Their thinking is very rigid.

I encourage you to find a community college course in any kind of trade discipline. It’s extremely rewarding to apply the lessons conceptually to your day job and life.

At University I asked if I could take a control theory class, and was told, "No, that's a dead field." Foolishly, I didn't. I learned more control theory from this one interview than anything in the last 40 years.

Funny. Control theory is vastly more necessary now than ever before. And there are people doing it; most of them just do it badly and call it AI.

It really does seem to have been a victim of (ironically) "schismogenesis" between the disciplines. The elephant has become so large people are only seeing parts of it. Off the top of my head I can see the following fragments:

A) cybernetics, in the machine sense: this is all op-amps and servoes, poles and zeroes. Taught as "control systems" in engineering courses.

B) organisational cybernetics: theory of management. If theoretical, this belongs in anthropology. Otherwise it's ended up in a cargo-cult form in MBA courses.

C) economic systems. Left to those economists who actually bother to do some differential equations rather than just writing thousand-word sponsored posts on why inequality is natural, or making catastrophic mistakes in Excel. Nobody's trying gigantic Stafford Beer systems any more.

D) Psychology and psychiatry: control and regulation seems to have become deeply unfashionable to talk about here, possibly due to replacing talk with drugs.

E) The literary wing: cyberpunk. Started as social commentary, but has faded into an aesthetic; nonetheless this is potentially the most powerful one for getting cybernetic thought into general circulation.

G) Environmental control systems: both the dull and dreadful world of IPCC reports on "forcings" and the more fanciful Gaia hypothesis inform us of how we change the environment and the environment changes us.

H) Advertising, gamification, and "dark UX": how can we use control systems and feedback loops through the human brain and emotional system to get people to put the maximum amount of coins in the slot?

The one thing these all have in common to me is the revelation of uncomfortable truths about ourselves, which probably accounts for their unpopularity. The original article talks about B and D with a bit of A and C, mostly pre-dating the others.

Reply on news.ycombinator.com