When we let programmers structure the world throug...
# linking-togetherj
Jimmy Miller
12/07/2021, 3:00 PM
When we let programmers structure the world through code, the strictures of the code's language frame what's possible, and therefore also the world they create and impose upon us
https://twitter.com/laurencediver/status/1468145966347923456
j
Jack Rusher
12/07/2021, 5:13 PMTbh, I don't think the problem with programmers structuring the world is the programming languages they use! 😊
k
Konrad Hinsen
12/07/2021, 5:36 PMThe author working in "computational law", I guess he considers any programming language (perhaps even any formal language) too constraining for structuring the world. And with that I'd agree.
j
Jimmy Miller
12/07/2021, 6:23 PM
Yeah, definitely not the only problem. But a small problem that is part of this larger problem.
Imagine if for example McCarthy's elephant 2000 existed. Now we have a language involving speech acts. A different conceptual framework than I think many of us programmers are familiar with.
As we structure the world more and more through code, we do have to ask what sorts of assumptions are pre-baked into our languages. What things are easy to express? What things are hard? Does that fact that most of our programming languages today are fairly similar have any impact on that?
👍 1
d
Denny Vrandečić
12/07/2021, 6:55 PMThe languages are rarely the problem (unless we go Sapir-Whorf here), but the problem is the models we express in those language and how we stick to them or not.
Put differently, the problem is not that your language has Booleans, it's when you use Boolean to model gender when you create a problem.
➕ 1
k
Konrad Hinsen
12/07/2021, 7:26 PMExactly. Inappropriate and/or premature formalization. Which happens not only in code. The numerical targets so loved by managers are quite similar.
j
Jimmy Miller
12/07/2021, 8:19 PM
Completely agree that models are the problem. I do think I'd just add to that that what models are easy or difficult to work with and express depends on our language.
Simple examples, Nats vs u32 vs integer vs floats. Some problems are better modeled with these various types. But it I'm in J's I'm probably modeling them with floats (the only real number type).
How many things do we not do or choose to not even consider because our languages make it difficult to express?
👍 1
e
Eric Gade
12/08/2021, 2:12 PMIf we compare computing systems to the actual concepts of language (natural) and writing, we can immediately see the shortcoming of talking solely about "programming languages"
Eric Gade
12/08/2021, 2:13 PMProgramming "languages" are not very good on their own and lack the most important feature of true natural language, which is that you get it for free no matter what
Eric Gade
12/08/2021, 2:14 PMIn this sense, programming languages are more like writing, which has to be learned over a sustained period of time, and which is the symbolic representation of language and linguistic thought.
Eric Gade
12/08/2021, 2:14 PMAll of this exists within an environment that gives it context and which enables its use, in standard or novel ways
Eric Gade
12/08/2021, 2:15 PMWhat I'm saying is that it's not enough to talk of programming languages on their own. It will get you nowhere. You have to talk about computing environments
Eric Gade
12/08/2021, 2:16 PMDoing most of your computing in a system that still uses a teletype metaphor at its core is going to be more restrictive to your potential mental models than whether or not language X uses floats as its number type
👍 1
s
shalabh
12/10/2021, 4:14 AMOn the topic of models vs reality: https://www.ppig.org/papers/2008-ppig-20th-blackwell/
The Abstract is ‘an Enemy’
👀 3