3 ms·

Saying programmers should "just do more work" is incredible economic naivety. Back in the day people dependencies were an obscure free software thing and every

Saying programmers should "just do more work" is incredible economic naivety. Back in the day people dependencies were an obscure free software thing and everyone just rewrote the same shit against Microsoft's or AT&T's interfaces, with maybe some in-house libraries. People sucked at making and using abstractions, and didn't do it. Then the pendulum swung and people started being download all sorts of crap and everything began to depend on everything else. Every once and a while the pile caves in and people star all over. People sucked at making and using abstractions, but did it. There's no way to roll back the clock. Forgoing code reuse is too costly. But we could write better abstractions and think about what it means to maintain whole library ecosystems (planning the division of labor and separation of concerns) rather than just individual libraries.

jfoutz 5y ago
Ftfa > They can perform more operations in one second than a human could in one hundred years. Please identify the traffic lights. We suck at abstractions, but I think we’re still winning there.
- Ericson2314 5y ago
  I'm a bit confused about what you are saying. "Please identify where our abstractinos go wrong"? "Our abstractions are slowly getting better"? "Abstractions are bad but not for performance"?
  - jfoutz 5y ago
    Ah, it was a late night quip with the intention of undermining the article. Yes, computers are fast, but there's no underlying intention. Our abstractions aren't great, but I think our abstractions let us communicate intention in a way that computers are far far away from achieving. Show a webpage with "identify all the traffic lights" to a child and, it might take a little while (especially if they can't read!) but the human will build up abstraction and leverage prior knowledge and solve the problem. Show the same webpage to a supercomputer, and nothing will happen, ever. The supercomputer has access to all of the required information via the internet. Heck, the instructions are right there on the page. But there's no intention of solving the problem. I guess, I'd say our abstractions do get better over time, but it can take a long time. an obvious example is Aristotle vs Newton vs Einstein. Computers haven't even started down that road. There are some vague ideas, and there has been concrete progress in narrow scopes. I guess I had a somewhat emotional reaction to the criticism of "programs must be written for people to read, and only incidentally for machines to execute." Which lead me down a crazy tangent. Performance matters, but only insofar as it helps people do stuff. But ultimately that's a people problem not a computer problem. The notion that computers are fast doesn't matter very much. We can use them more efficiently if we want to, but do we really want to? Who cares? Maybe it does make sense, maybe it's better to build a new feature. Anyway, crazy tangent.
    - Ericson2314 5y ago
      > I guess I had a somewhat emotional reaction to the criticism... I agree wtith the rest of your post there. > I think our abstractions let us communicate intention in a way that computers are far far away from achieving. > > Show a webpage with "identify all the traffic lights" to a child and, it might take a little while (especially if they can't read!) but the human will build up abstraction and leverage prior knowledge and solve the problem. Ah. I don't measure progress using computers relative to the human brain. So when I say programmers are using bad abstractions, I am not envious of human pattern recognition etc. at all. Likewise natural language abstractions are messy and culturally loaded and so much more I don't want to get into. When I say abstractions I am just talking about the rigid formal ones we program. I think the field is too focused on A.I. stuff because of old romantic notions about human vs computer, piss poor education making programming knowledge scarce, and general laziness among the non-technical powers that be about defining the requirements. I don't think should be the main thread of how we use computers in society.

jfoutz 5y ago

Ftfa > They can perform more operations in one second than a human could in one hundred years. Please identify the traffic lights. We suck at abstractions, but I think we’re still winning there.

Ericson2314 5y ago
I'm a bit confused about what you are saying. "Please identify where our abstractinos go wrong"? "Our abstractions are slowly getting better"? "Abstractions are bad but not for performance"?
- jfoutz 5y ago
  Ah, it was a late night quip with the intention of undermining the article. Yes, computers are fast, but there's no underlying intention. Our abstractions aren't great, but I think our abstractions let us communicate intention in a way that computers are far far away from achieving. Show a webpage with "identify all the traffic lights" to a child and, it might take a little while (especially if they can't read!) but the human will build up abstraction and leverage prior knowledge and solve the problem. Show the same webpage to a supercomputer, and nothing will happen, ever. The supercomputer has access to all of the required information via the internet. Heck, the instructions are right there on the page. But there's no intention of solving the problem. I guess, I'd say our abstractions do get better over time, but it can take a long time. an obvious example is Aristotle vs Newton vs Einstein. Computers haven't even started down that road. There are some vague ideas, and there has been concrete progress in narrow scopes. I guess I had a somewhat emotional reaction to the criticism of "programs must be written for people to read, and only incidentally for machines to execute." Which lead me down a crazy tangent. Performance matters, but only insofar as it helps people do stuff. But ultimately that's a people problem not a computer problem. The notion that computers are fast doesn't matter very much. We can use them more efficiently if we want to, but do we really want to? Who cares? Maybe it does make sense, maybe it's better to build a new feature. Anyway, crazy tangent.
  - Ericson2314 5y ago
    > I guess I had a somewhat emotional reaction to the criticism... I agree wtith the rest of your post there. > I think our abstractions let us communicate intention in a way that computers are far far away from achieving. > > Show a webpage with "identify all the traffic lights" to a child and, it might take a little while (especially if they can't read!) but the human will build up abstraction and leverage prior knowledge and solve the problem. Ah. I don't measure progress using computers relative to the human brain. So when I say programmers are using bad abstractions, I am not envious of human pattern recognition etc. at all. Likewise natural language abstractions are messy and culturally loaded and so much more I don't want to get into. When I say abstractions I am just talking about the rigid formal ones we program. I think the field is too focused on A.I. stuff because of old romantic notions about human vs computer, piss poor education making programming knowledge scarce, and general laziness among the non-technical powers that be about defining the requirements. I don't think should be the main thread of how we use computers in society.