The first several hours of the contest should be more focused on developing interfaces (also voice conferences suck)

[fj128]

Voice conferencing sucked this time. It was, like, "oh we should make interfaces". Who are the "we" was left entirely unspecified, because of course nobody had (or could possibly have) the power to say "you dude, go write and commit so and so interface".

In my ideal world it would be the chat model where someone who has finished reading the specification says "I'm going to implement the commands.py" and someone else says "I'm going to implement data_files.py" and someone says "I'm going to implement emulator.py", and that's what happens.

The voice conferencing instead quickly devolved into the worst thing about these contests, where people speculate about solvers without having a single clue about the problem.

---

The important thing here though: implement the interface first, commit and push, and Request For Comments. This was a bad thing that I did not do, I should have pushed the commands.py very fast and then try to implement the parsing/composing logic (which took several hours) later.

When I did push the interface the RFC part was good, Vlad telling me that in his code the difference between Pos and Diff was important so I fixed my code to use those.

I'm pretty sure that if we had unimplemented interfaces for querying the world state for can_apply(command : Command) at around +2 hours (instead of +48 hours), we'd have much more people writing solvers. And then we'd implement those interfaces and start running those solvers and everything would have been awesome instead of totally demoralizing.

[earthdok]

+1

[fj128]

Oh, and by the way, in 2017 we actually had productive voice conferences later on IIRC, so not only this one sucked all by itself because of being totally premature, it also prevented further possible good voice conferences.