The Only Winning Move - Loris Cro

Software You Can Love Vancouver 2023

6 Likes

Great talk @kristoff!

3 Likes

Interesting talk, @kristoff!

One thing in particular stood out for me: the point about building utilities that you ended up not using. One way to get around this is to stop writing software and start thinking about what you actually want (training that non-dominant arm via your example).

For about 5 years, I’ve worked on an assistant AI program that has changed shape over time. I wanted the ability to ask arithmetic questions and have it return correct responses. So… I agonized for a while about how to do this. I wrote a parser, trained dispatch for utility functions and… it was subpar. Eventually, I got approval to use wolfram alpha with my app and opted to detect those questions, send a packet over the network, and unbox the response. It worked! I was pleased and felt like it was a big deal at the time.

Eventually, I realized that I absolutely never used it whatsoever. Why? Because using a calculator was easier. All the work that went into proving that point ended up being way less effective than using the basic utilities that I’ve always had.

In other words, to write better software that actually suited my needs, I had to stop writing and start thinking about things outside of the problem itself.

4 Likes

I think this is paradoxically a very important step in building software and at the same time one of the hardest; having the discipline and willpower to stop coding and think. Also, this applies to many sub-levels of software development, for example, features within a project. You may have decided to make an actually useful tool, but then start piling on cool features that aren’t really necessary and thus probably never be used.

3 Likes

In my case specifically, when I wrote the ticketing system for the original software you can love I think I did a decent job with most things but I ended up with some silly parts that in the end made not too much sense. This is something that one can only improve with first hand experience and in fact I’m working on a new version as we speak for Software You Can Love 2024.

4 Likes