[NTLK] Einstein, last call
Eckhart Köppen
eck at 40hz.org
Fri Oct 23 10:46:25 PDT 2026
On Fri, 23 Oct 2026 19:24:56 +0200, Matthias Melcher wrote:
> Next, I ran Claude over all the software I wrote in my life. It gave
> me a lit of bugs that it found with an hour, and asked if it should
> fix them, all of them, in probably just another hour.
>
> This is just insane.
Yep, it is. It has a bunch of interesting implications to the way we
implement and treat software: Software can be created very fast now,
the act of implementation is really secondary. We can create whatever
is needed, and we don't even need to share this, this is the "OSS is
dead" argument, and I think there is something to it. If I need a
NewtonOS emulator, I can just create one (as long as I have enough
tokens).
I've created a lot of tools over the past year which would potentially
be useful for others, but I don't feel like releasing them because I
want the freedom to do what I want with them, and others can create
these tools just the same. It's a bit sad actually :)
Now if we have a project like newton-re or Einstein, which is huge,
beneficial and useful to many users, the question is how do we run
these projects? Pull requests are almost too fine grained... I'm
starting to lean towards "requirements based software engineering" as
in instead of code, I agree with the project maintainers on a new or
changed requirement, put that into a pull request and let AI implement
it.
I've had that discussion in other projects, where writing a good bug
report was as effective as pushing code. The bug report is even more
valuable since that would anyway need to be reviewed and agreed upon,
same with requirements. A pull request might nowadays be even a step
too far, it potentially shortcuts the discussion why it came to life.
Thus the policy I've seen now more and more of no pull requests without
a linked issue.
The software engineer in me is very very happy about this, because is
shifts the effort to producing clear requirements (and I blame the
agile movement to have screwed that up ;). The programmer in me is
unhappy, but is coming to terms with the new reality.
Eckhart
More information about the NewtonTalk
mailing list