Conversation

Notices

1. Embed this notice
   clacke (clacke@libranet.de)'s status on Wednesday, 30-Apr-2025 10:22:30 JST clacke
   It's interesting to think about what we mean when we say "in theory ... but in practice".
   • Embed this notice
     Fish of Rage (sun@shitposter.world)'s status on Wednesday, 30-Apr-2025 10:22:29 JST Fish of Rage

     in reply to
     @clacke "In theory, there is no difference between practice and theory. In practice, there is." -- Yogi Berra
     clacke likes this.
   • Embed this notice
     clacke (clacke@libranet.de)'s status on Wednesday, 30-Apr-2025 12:44:17 JST clacke

     in reply to

     As the funniest saying of them goes "in theory, theory and practice are the same; in practice, they're not".

     What do we mean by that?

   • Embed this notice
     clacke (clacke@libranet.de)'s status on Wednesday, 30-Apr-2025 12:44:24 JST clacke

     in reply to
     No matter what the mismatch is between practice and theory, of course practice is what matters in the end. If theory doesn't match practice, that's usually because practice is messy and complex, and theory needs to be reduced to a level of complexity that actually helps us think.
   • Embed this notice
     clacke (clacke@libranet.de)'s status on Wednesday, 30-Apr-2025 12:44:31 JST clacke

     in reply to

     • Fish of Rage
     @sun libranet.de/display/0b6b25a8-1… 😁
     Fish of Rage likes this.
   • Embed this notice
     :copyleft: Kiri :tux: (kiri@fosstodon.org)'s status on Wednesday, 30-Apr-2025 12:44:34 JST :copyleft: Kiri :tux:

     in reply to

     @clacke
     I've always subscribed to the notion of "If theory doesn't match practice, then it's because it doesn't match, but there is a right theory that does match"

     clacke likes this.
   • Embed this notice
     clacke (clacke@libranet.de)'s status on Wednesday, 30-Apr-2025 12:44:35 JST clacke

     in reply to

     • :copyleft: Kiri :tux:
     @kiri Ah, but is that theory helpful? Maybe we don't use it because it's unwieldy. A 100% accurate theory basically comes down to emulating every particle and force in the universe.
   • Embed this notice
     clacke (clacke@libranet.de)'s status on Wednesday, 30-Apr-2025 12:44:35 JST clacke

     in reply to
     When we say "it works in theory, but it'll never work in practice", that suggests that the theory is leaving something out, something significant. That's kind of obvious.
   • Embed this notice
     clacke (clacke@libranet.de)'s status on Wednesday, 30-Apr-2025 12:44:36 JST clacke

     in reply to

     The more interesting case is when it doesn't work in theory.

     Sometimes people misunderstand this statement, but for people used to thinking in the abstract and formulate models, it means basically this: "there is an unavoidable tradeoff that any solution needs to make, but success requires that we don't make that tradeoff, that we can have both A and B"

     If it doesn't even work in theory, it has no chance, none, of working in practice. You can't just do it harder, better, faster or stronger, because you're running into some unbreakable limit.

   • Embed this notice
     clacke (clacke@libranet.de)'s status on Wednesday, 30-Apr-2025 12:44:37 JST clacke

     in reply to

     Sometimes this is said, and then someone goes and does it in practice anyway. Usually that means they redefined "it". The problem statement was simplified or too strict, and it turns out that something that the actual thing we needed in practice *was* possible.

     You can't solve an NP problem in polynomial time, in theory or in practice, but maybe you turned around and solved a different problem, usually by replacing the "correct" answer, which required an NP-complex solution, with a good enough answer, which you could get with a P-complex heuristic.

   • Embed this notice
     clacke (clacke@libranet.de)'s status on Wednesday, 30-Apr-2025 12:44:37 JST clacke

     in reply to

     Or, you know, again, just like when the theory said it was possible, but in practice it wasn't, the theory was just incomplete and failed to include some dimension.

     It is not "possible" for one process to rewrite the memory of another process, but using the RowHammer exploit, which doesn't exist in the usual memory and process model, but exists in the practical CPU, you can do it.

     This is an interesting case, because unlike a natural science model that tries to calculate reality and fails, like the Newtonian model failing to predict Mercury's motions, here the model comes first, and it is the hardware.that fails to implement it in practice.

   • Embed this notice
     clacke (clacke@libranet.de)'s status on Wednesday, 30-Apr-2025 12:44:50 JST clacke

     in reply to
     Maybe you just get angry at the CPU manufacturer for failing to implement the model cleanly ... but if that CPU is all you have and you want your process isolation to work, you'd better include that flaw in your model.
   • Embed this notice
     :copyleft: Kiri :tux: (kiri@fosstodon.org)'s status on Friday, 02-May-2025 00:43:30 JST :copyleft: Kiri :tux:

     in reply to

     @clacke
     Well it doesn't have to be perfectly coupled lol, some margin of error will always exist

     clacke likes this.
   • Embed this notice
     Mans R (mansr@society.oftrolls.com)'s status on Friday, 02-May-2025 00:44:27 JST Mans R

     in reply to

     @clacke It can also mean that something is possible but very difficult. Like riding a unicycle on a slack rope. A few people can do it, but if it's the only option for crossing a river, most of us are better off not trying.

     clacke likes this.