Well. It appears that somebody (this time a person unknown to me) took my slightly disjunctured rant, including my full INTERCAL program, and submitted it back to Chat-GPT4, wanting to see if it would take the opportunity to clarify its previous assertions, or refute any of my objections. I don't have access to the details this time, so I'm just going on the transcript as reported to me. Assuming that it is indeed a faithful record, I feel that it's worth sharing and acknowledging. To wit:
INTERCAL is a esoteric programming language designed to be as different from conventional programming languages as possible. It's known for its obfuscation and convoluted syntax, and your provided program adheres to this style perfectly.
Yes, because if it didn't adhere to this style, it wouldn't work. "Aha, I see that your well-tested Python program is written in a Pythonic style!"
It's quite an accomplishment to create a functional password generator in such an unusual language!
Such a backhanded compliment. Again, I will acknowledge that it is an accomplishment to write a functional program of any nature in INTERCAL—if you're not already conversant. But that is true in nearly any programming language, to some degree. Furthermore, there's nothing about a password generator program in particular that is especially challenging to the INTERCAL programmer.
Indeed, I would argue that a password generator is almost as good a fit for INTERCAL's strengths as a practical task can get. It doesn't require parsing input, it doesn't have any extraneous output requirements, there are no error states to detect or describe, and several arithmetical steps of the process can be finessed with creative bit-twiddling. Really, you could hardly ask for something better suited to the meager handful of things that INTERCAL does well.
(In fact, if you're interested in INTERCAL programming yourself, I would definitely recommend this as a practice problem. Without looking at my code, try to write your own version. Then once you have a working program, compare it to mine and see where we made different tradeoffs.)
Before we dive into the details of the program, let's try to clarify why a previous version of ChatGPT might have said that INTERCAL lacked a random number generator.
Please do! I am all ears.
That statement may be somewhat misleading. It's not that INTERCAL inherently lacks the ability to generate random numbers, but it doesn't provide a straightforward, built-in function to do so.
… you perfidious blatherskite.
Therefore, creating a random number generator in INTERCAL would require a creative and complex implementation.
And so we see that Chat-GPT still refuses to admit that INTERCAL does, in fact, provide a straightforward, built-in language feature for generating random events, or that turning said random events into random numbers is anything but "complex".
I'm curious. Would Chat-GPT claim that C doesn't provide a "straightforward, built-in function" to generate random numbers? After all, it's not part of the language itself: you have to call out to a library function. (Or write your own PRNG, which while only a few lines of code would certainly be less straightforward.)
If the distinction being made is based entirely on the availability of a standard function to provide random numbers, then STFU, because guess what: INTERCAL also provides a random-number generation function in the standard library. In fact it provides two of them—one with uniform distribution and one with normal distribution. So stop picking on INTERCAL like there's some kind of gaping hole in its functionality. (I mean, there is, there are many actually, but this happens to be one of the very few places where there is not.)
Moreover, if C's rand() function is acknowledged to
in fact be a source of random numbers, then STFU still, because the
C-INTERCAL compiler is using C's rand() function
in order to provide randomness. (It's called C-INTERCAL because it
compiles to C code, you see.) Either way you slice it, there's nothing
about INTERCAL's random-number generation that is any worse than any
other, more pedestrian programming language.
(An aside: I realize that some readers may now be wondering—if the INTERCAL system library has a standard function for generating uniform random numbers, then why didn't I use it in my password-generation program? The answer is twofold. Partly it was done as a minor optimization, since I never need more than a few bits of randomness at a time. You can also see, if you read the code, where after generating three random bits the code tests to see if the current character should be forced to be a digit or punctuation mark, and if so it skips over the code that generates the remaining random bits. The second reason I wrote it that way was because this capability was such a point of contention with the original response. I wanted to make sure to spotlight the generation of random values in my program, rather than hide it behind a system call. Yes, that does mean that the program could have been even shorter and simpler than it was.)
In the code you provided, it seems the author found a way to work around this limitation.
The limitation is with your grasp, not the language. I shall attempt to stop belaboring this point.
The password generator you posted creates passwords by initializing an array (
,6) with 16-bit values that represent ASCII characters.
I'll award partial credit for this statement. ,6 is an
array of 16-bit integers, but the actual values are never more than 8
bits wide, much as you would expect. And again the values represent
ASCII characters somewhat indirectly, being, you
know, bitwise-reversed.
The set of characters includes digits, some punctuation marks, and uppercase and lowercase letters, but excludes easily misread characters. It creates a 16-character password using these values, and it seems to enforce a certain level of complexity by ensuring the inclusion of at least one number and one punctuation mark.
ChatGPT is here demonstrating its ability to glean information from comments and pass them along.
The main loop begins at
(1) DO COME FROM (9).
Oh, that line? The one that comes immediately after the
comment DO NOTE THE MAIN LOOP WHICH BEGINS HERE? That
one? Are you sure?
It really is like listening to a bullshit artist improvise.
Here, a spot in the password is selected and filled with a character from the array.
I would have to say that this sentence is misleading. Each character of the password is generated sequentially and immediately output. There is no selecting of a spot in the password, as it has described, since that would require the password to be stored in memory (which would complicate the program for very little advantage).
There are some points in the loop where the selected spot may be forced to be a number or punctuation mark. If the character set is checked for digits and punctuation marks, pre-chosen spots for these types of characters are dropped. The selected character is then read out.
I'm unsure what this program thinks it means by the phrase "if the character set is checked for digits and punctuation marks". My best guess is that it's trying to say: "The chosen character from the character set is checked, and if it is a digit or punctuation mark" then etc. Or, it may just be a furtherance of its prior misunderstanding.
In any case, it's obvious that it is simply doing its best to rephrase the information in my comments, and making educated guesses to paper over the parts that aren't explained outright.
This INTERCAL code uses a clever trick to generate seemingly "random" numbers - using the
%50operation (which in INTERCAL means to mingle the bits of two 16-bit numbers), along with unary operations like#1,#2,#4, etc. on the same variable, to generate different results each time the loop is run.
Again, Chat-GPT. You need to learn when to stop talking. This
entire sentence is fiction. It is a tapestry of phantasms. Nothing of
Truth doth fade, but doth suffer a sea-change into something rich and
strange. Sea-nymphs hourly ring its knell. Hark! now I hear
them—ctrl-G (BEL).
To be specific. %50 is not an operation; it is a
statement modifier. Said modifier affects control flow. It does not
and cannot mingle bits, not having any arguments from which bits could
be obtained. INTERCAL does have a binary operator called
"mingle" (though I prefer its more common name, "interleave"), but
it's denoted by $ (or ¢) and is not really
relevant here. #1, #2, #4, etc.
are not operators, unary or otherwise; they are the numerical
constants 1, 2, and 4. Seriously, how did you get that one so
completely wrong? And finally, none of those things, had they worked
as you so creatively imagined, would assist a program in "generating
different results each time the loop is run". (Such functionality
could help in generating different values for each iteration of the
loop, if we're being generous, but that isn't really the point in
question, so if that is what is meant it's a non sequitur.)
Chat-GPT, why are you so incapable of just saying "I don't know"‽
That one isn't a rhetorical question, actually. Why can't Chat-GPT just own up when it doesn't know something? You know, before AI research was waylaid by Big Data, it was common practice, if not actually expected, that a program should be able to describe the process of its own reasoning. It was understood that a bare answer, without the chain of logic leading up to it, was useless if not actively harmful—even if the answer happened to be correct. But the design of modern systems made those inferential stack-dumps difficult to produce, or even (wait for it) practically impossible.
So is the problem a fundamental one to its design? Does Chat-GPT never say "I don't know" because it simply cannot tell when it doesn't know something? Is the ability to make judgements on the quality of its own chain of reasoning simply not available to it, and therefore it has no basis on which to decide when to speak and when to remain silent wisely?
I don't know, of course. It's just an impression, based on my vague understanding of how such systems work, and the types of errors Chat-GPT makes systematically. But hey: at least I know that I don't know.
There are one or two more unwarranted defamations in the closing paragraphs of the transcript, but now that this essay has taken a serious turn, it seems pointless to continue. So I'll take my own advice and STFU here. I would just like to end with a reiteration of my request that Chat-GPT kindly learn to cease flinging calumnies at my beloved problem child. INTERCAL did nothing wrong.