Replying to @LewisCTech @sudo_goreng
This is definitely one topic we disagree at a fundamental level.
2
Replying to @anneapplebaum
The admiration for Putin is largely driven by his and his administration's anti-EU ideology. They see the EU as a threat to American economic interests.
He's a scoundrel of the highest order, but he's far from alone. Putin is merely a pawn.
9
Replying to @valigo
False equivalences are an hobby, it seems.
Comparing lines of code to mathematical formulations another.
And a project with over four thousands contributors, with the last merge 12 minutes ago, with an single mathematician AI generated proof, yet another.
7
1,341
My last post... promise! So, I've always been skeptical of formal methods in SWE. I believe they can have a place in small parts of a full system, maybe. Or in specialized projects.
Unless, of course, we believe LLMs will do it and we don't ever need to validate any of it. Yup.
8
And the pressure this puts on the requirements phase come with one additional problem: The unsolvable problem of clients clearly defining their needs.
Because you can't your cake. No, no! If we want formal validation, we must accept we need to write true propositions.
1
6
But in the world of distributed computing, business and general-purpose software, this adds an immense amount of upfront and maintenance costs to the development pipeline. I could go in more detail why I think so, if you want. But I feel you know what I mean.
1
1
4
Replying to @LewisCTech @sudo_goreng
So, I skimmed through the paper. Don't get me wrong, I think it is indeed fascinating, But I have no motivation to study it in detail.
It is clear to me this is only of any significance to highly specialized software with highly specialized teams.
2
5
Replying to @LewisCTech @sudo_goreng
Also behavioral scientists and psychologists to help translate clients requests into actual true statements of the environmental requests
8
Replying to @sudo_goreng
And don't forget the fun part: keeping up with requirements changes and the fast-pace addition of new requirements.
Who said mathematicians were out of job? Modern companies will need armies.
35
Replying to @LewisCTech
If you pair that with a fedora, I swear I'll come over for beers and demand pictures with my kids!
1
1
6
I'm starting to think that our collective understanding of science reduces as time moves forward. I've never been more confident of this.
Replying to @JustDeezGuy
It would require me to go and cherry pick. Not interested, at all!
But if you believe that progress in mathematics and physics has been done by people feeding Lean machines with their proofs and publishing them as authoritative work not for review, I don't know what to tell you.
21
Replying to @LewisCTech
I bookmarked it, but haven't got yet the time to give it the attention it deserves. I'm aware from past conversations what the subject matter is all about and have been waiting for it.
As for HN... you wanted to be a public figure...
1
3
42
Replying to @JustDeezGuy
It would require me to go and cherry pick. Not interested, at all!
But if you believe that progress in mathematics and physics has been done by people feeding Lean machines with their proofs and publishing them as authoritative work not for review, I don't know what to tell you.
Replying to @JustDeezGuy
The distinction I'm making is fundamental in scientific research and has been the modus operandi since the first formal systems were developed.
If you can identify where I am making an invalid assertion, say so. Otherwise...
68
Replying to @fromdevoid @JustDeezGuy
Mathematicians have been, as you certainly know, some of the main users of Lean kernels. And they have certainly faced many situations in which a verified proof turns out to be invalid.
So, no. It's not the proof. It's a proof that now needs to be validated. All 175k files of it.
1
28
Replying to @ptrschmdtnlsn @JustDeezGuy
You should maybe stop doing that and instead engage with the argument itself, or be quite.
"AI generated" counter-arguments are rapidly becoming the biggest straw man on the internet. Are you into it?
1
59
Replying to @JustDeezGuy
The kernels checks that something is a valid proof of the stated propositions. It does not establish that the propositions correspond to the truth, or that the assumptions themselves are true.
You still need formal validation to establish the proof. That's peer review.
2
252

