Dan Luu
patreon
Baseball scouting reports
🕑 Added 2019-05-18 13:00:02 +0000 UTCI find baseball scouting reports from before stats were "really" used in baseball fun to read. Like any major sport, productivity is much easier to quantify than in most real-world endeavors, like programming since the game is much simpler than "real" problems are. And, among major U.S. sports, baseball is the easiest sport to quantify.
Even though many millions of dollars were on the line, rather than attempt to do this, teams decided who to draft based on scouting reports containing things like the following.
... will be a real specimen with chance to have a Dave Parker body. Facially looks like Leon Wagner. Good body flexibility. Very large hands.
Jim Abbott:
Outstanding physical specimen -- big athletic frame with broad shoulders and long, solid arms and leg. Good bounce in his step and above avg body control. Good strong face.
Derek Jeter:
Hi butt, longish arms & legs, leanish torso, young colt
A different scout on Jeter:
Wiry loose good agility with good face
Lots of comments about how someone has a "good face", who they look like, what their butt looks like, etc. Out of context, you might think they were scouting actors or models.
For a couple decades, stats nerds (not professional statisticians, just baseball fans who thought that you could apply stats to baseball), used stats 101 techniques to make predictions about how good players would be and wiped the floor with the professional establishment, which was going on data like "Facially looks like Leon Wagner" or "Hi [sic] butt".
Hiring in baseball eventually came around and teams now employ professional statisticians to figure out how productive players are. I don't know that you can really do the same thing for hiring in programming, but I find it strange that basically no one even tries. I only know of one company that's made any attempt at quantifying how well their hiring process works, and from what I've heard, the methodology isn't very good.
There are a couple of companies that are trying to fix hiring and write blog posts about using data to inform hiring decision, but those companies aren't really tech companies and the data analyses they've published don't seem really sound. It appears to me that we're not even doing hiring as well as baseball teams were during the "good strong face" era of baseball hiring. For us to be doing that well, there'd have to be data-driven discussions of hiring. The closest thing I've seen is https://sockpuppet.org/blog/2015/03/06/the-hiring-post/, and while that post is reasonable, there doesn't seem to be a community that's taking this idea and going anywhere with it.
