[FRIAM] naive question
Marcus Daniels
marcus at snoutfarm.com
Fri Oct 21 13:35:47 EDT 2022
The toy code problem is far more severe in the quantum computing (QC) world. In QC there are hundreds of papers that consider systems of less than 10 variables and claim great results. When modern optimization and ML operate at the scale of millions of variables.
-----Original Message-----
From: Friam <friam-bounces at redfish.com> On Behalf Of glen
Sent: Friday, October 21, 2022 10:19 AM
To: friam at redfish.com
Subject: Re: [FRIAM] naive question
Ha! If we're going to argue about words, then let's stick with the word "identity" and skip the "metaphor" nonsense. You and Frank seem to be using the word in a weird way. Identity means "the exact same particular thing over any differencing available" or somesuch. I mean, it's used that way in phrases like "identity theft" as well as mathematical identity. It's equivalence sets all the way down. I just can't imagine any working computationalist would ever say anything like "executed identically" unless ... well ... the exact same process, with the exact same steps, happened.
I suppose there are deep philosophical intuitions pried at by the words "emulation" versus "simulation". And one can argue (again with help from Christian List) about whether there exist fully closed ontological walls like we try to create with things like Jails, HyperV, Docker, VM's like Java's, etc. But "execute identically" is a phrase that would only be used by someone who worked *way* above such levels (assuming levels even exist at all). It's a bit like talking to the kids programming websites these days, with access to infinite disk space, infinite memory, steeped in continuous delivery, etc. [⛧]
Layers of abstraction are fine. Use 'em when you need 'em. But we shouldn't posture by invoking things like "instruction sets" and "execute identically" in the same breath. (Not that you did that ... just sayin'.)
[⛧] Rant: This is a good talk <https://www.youtube.com/watch?v=8Ab3ArE8W3s>. But I get super irritated when people use *toy* code in their rhetoric and leave large scale deployment as an exercise for the reader. Yeah, fine. The REPL is cool and all. But when my simulation takes a fvcking WEEK to execute, it's difficult to sympathize. I've recently been playing around with VSCodium, which is pretty cool. But whatever, man. I still have to upload the code somewhere and execute it. Get off my lawn!
On 10/21/22 09:24, Steve Smith wrote:
>
> As a counter-example, we ran film recorders whose "guts" were built by Ed Fredkin's Information International company and were built to the spec of Dec PDP-11 (I think 11?) and it was anecdotally agreed among the user community (of a few thousand delivered units in the world?) that these PDP-clones *never* failed to execute the code identically to the machines they were patterned after. I don't remember the details of implementation of these 70's era hardware designs, but I understood that they III designed their own PCBs but (obviously?) used the same CPU chips... I don't know about all the other support components... A likely answer to this pondering is that these machines did not run a general purpose OS and the III software/system people probably made up for any differences in Software/Timing/Error Handling?
>
> If Owen is listening in here, I think he was there for more than a little of this from inside Apple/Sun?
>
> - Steve
>
> PS. To concede/confront glen's sentiment that: " 'Metaphor' is an evil word, used only by manipulators and gaslighters", I would offer that the use of *conceptual metaphor* is to thinking as noise is to simulated annealing, and his point about "tighter or looser equivalence" might well be the best argument *for* the use of metaphorical thinking? I can't believe I'm stirring/kicking this can of worm-hornets down the street again...
>
--
ꙮ Mɥǝu ǝlǝdɥɐuʇs ɟᴉƃɥʇ' ʇɥǝ ƃɹɐss snɟɟǝɹs˙ ꙮ
-. --- - / ...- .- .-.. .. -.. / -- --- .-. ... . / -.-. --- -.. .
FRIAM Applied Complexity Group listserv
Fridays 9a-12p Friday St. Johns Cafe / Thursdays 9a-12p Zoom https://bit.ly/virtualfriam
to (un)subscribe http://redfish.com/mailman/listinfo/friam_redfish.com
FRIAM-COMIC http://friam-comic.blogspot.com/
archives: 5/2017 thru present https://redfish.com/pipermail/friam_redfish.com/
1/2003 thru 6/2021 http://friam.383.s1.nabble.com/
More information about the Friam
mailing list