I think a group of 15-20 bright and enthusiastic CS students under the guidance of a tenured CS professor could accomplish a lot with the time/effort they were spending and learn just as much!
* Introduction of developers to a core kit of useful algorithms
* Practical examples for developers to design algorithms that require adaptation (e.g., how to adapt the bipartite matching algorithm to tripartite)
* Training people in how to debug their code--the most useful competitions are the ones where programmers don't always have access to the computer, so you have to be able to step through your code without running it
At the same time, however, I think there is rapidly diminishing returns here. You'll get a lot of the first bit of effort you put into programming competitions, but then the improvement you get is incredibly incremental. I'm especially unimpressed by race-to-the-finish kind of competitions, since I think the skills it requires to get very high on the leaderboards are not useful.
The first two have published some influential works during their academic careers:
https://scholar.google.com/citations?user=df8TSy4AAAAJ
https://scholar.google.com/citations?user=Vjo4Tg4AAAAJ
I wonder what the total number of research papers and citations of all Code Jam winners would be. Sadly, it's hard to find reliable info on most of them.
Still, I think it's pretty sad that "results don't conform to our ideological results" equates to "it must be a bad metric, let's shut it down!" Hope it's not the same kind of thinking that led to ElectronConf being cancelled a couple years ago for "not having a diverse slate of speakers", despite the fact that all proposals where chosen in a blinded process explicitly set up in a way to eliminate bias: https://news.ycombinator.com/item?id=14480868
<https://www.theregister.com/2023/01/27/google_open_source/>
Steven's been around the community a long time and is pretty well connected, so may know more than he'd stated explicitly in that piece.
All that said, by shutting down without announcing a reason, Google must've known people would speculate and likely assume it's because of the "Google = Oracle" reasoning. So I'm certainly leaning towards the "this is pretty sad" side of things.