0 and 1 are physically different from each other. A 64-bit float could easily be seen as an array of bytes without any change to the information it contains.
Bottom line: types are external to the information something contains.
What tells you that a particular location in a chip represents a bit and what voltages correspond to 0 and 1? You have to bring external knowledge of the particular device.
Also through this I can discover which voltage corresponds to the high and low state.
Meanwhile a series of bytes can be a float, a string, a table of numbers or even a function pointer - you can't tell which without external information.
I think what you're doing is taking your existing understanding of how a processor works, and saying that you could figure it out, given an actual processor. But that's begging the question. You have a definition for what a "bit" is, and what a "byte" is, and since you know what they are you can go on a hunt within a physical system and label things as "bit" and "byte". But I could come in and start labeling certain things "function pointer" or "float", and I don't see a particularly compelling reason why that's not allowed, but labeling things as "bit" or "byte" is allowed.
Or in other words, why is, "I know how synchronous digital circuits work and I can figure out a processor" acceptable, but "I know how compilers work and I can reverse engineer a program" prohibited?
And I'll still say that the statement that, "Typed programming languages don’t exist," is not even wrong. It's just a cute thing that you can say, to demonstrate that you can choose a definition for "exist" or "typed programming languages" that makes it needlessly difficult to have a conversation.
We agree that the stuff at the bottom exists, if the word "exists" has any meaning at all. But if you argue that "things which are just conventions about how to interpret other things" (like types) are not real and do not exist, if you want to be consistent about that, the entire stack evaporates and you are left with whatever is at the bottom, such as fundamental particles.
For a more concrete example, you said that, "A type is just a convention about an agreed interpretation of those bits." The major problem with this statement is that a program is also an agreed interpretation of some bits. Whether some piece of memory is an instruction or data is just a matter of interpretation. So if this argument says that types do not exist at runtime, then programs do not exist either.
So if we accept that programs exist, the argument must be wrong.
I don’t understand this part. Where do we agree that? As opposed to those things just continuing to be useful fictions at the boundary between the current best-known reductive explanation of a thing and the outstanding questions where that reductive explanation fails to reproduce observational data as-yet-unexplained?
We agree that there is a physical, objective universe. (Why? We assume it is true because otherwise conversation is not possible.) This physical universe exists (in a physical sense). We also have a catalogue of other phenomena (silicon atom, apple, Kentucky) which we say "exist", phenomena which don't exist (magnetic monopole, dragon, The Shire), and a catalogue of intangible concepts (happiness, vector space, triangle) which don't exist in a physical sense.
But every term here, except "the physical universe," are either concepts or interpretations of phenomena. If you say that a data type does not exist because it is nothing more than an interpretation of a physical phenomenon, then my problem with this is that this also applies to computer programs, bits, bytes, silicon atoms, apples, and Kentucky. So I would prefer to use a definition for "exist" that includes data types, bits, bytes, silicon atoms, apples, and Kentucky.
I don't think there's a compelling case for drawing a line between data types and bits and saying "bits exist, but data types do not".
I said that I didn’t understand your argument, so I asked some questions. And then I expanded my argument with reasoning, but apparently you feel that my reasoning is deficient, but lazily assert that I did not offer “any reason” to believe them. I honestly thought that I did offer reasons.
> In fact I’d say your whole series of comments here are rhetorical gimmick with some bravado like it’s a great philosophical insight while side-stepping the only interesting philosophical question here, which is how do we decide which concepts “actually” exist.
That’s an interesting philosophical question, but we’re having a discussion about programming languages and type systems, and I would say that the philosophical question of “what exists?” is side-stepping the interesting questions about type systems. In order to have a discussion about type systems we need to agree about what exists, and the best way to do that is to simply agree by convention that certain things (like type systems) exist.
Which is what I’d prefer. Same thing with mathematics… if we are talking about vector spaces, I don’t want to have a discussion about what “exist” means, I just want to use the word “exist” as a substitute for mathematical concepts, agree how we use the word in context, and then talk about math.