I've seen this a bunch in my timeline yesterday and today. As the person running Cornell CS PhD admissions this year, I have Thoughts (some of which have been stewing for longer than the past day).https://twitter.com/andreas_madsen/status/1211329218619092993 …
-
-
There is not a secret sauce here. It is a multi-objective problem, and we go by the signals we see across the entirety of the folder. For faculty candidates, we ask for 1-2 top papers (CRA best practices memo). So why count pubs for PhD applicants?
Prikaži ovu nit -
Third: While we do look for research experience (a PhD is a heckuva slog to commit to without a prior taste), I find the idea of undergrads hearing "publish or perish" to be terrible -- and the hyper-focus on a few venues even worse.
Prikaži ovu nit -
We have this ideal of the "T-shaped" PhD graduate: competent across a range of things (the top of the T), deep in one (the stem of the T). I think the researchers I find most interesting tend to branch off from there.
Prikaži ovu nit -
Finish a PhD, and there is still a career of 3-4 decades ahead. For a lot of people who go into research, that means 3-4 decades of moving between different projects and interests. How does immediate hyper-specialization help with that?
Prikaži ovu nit -
Fourth: There's more to the world than machine learning! I assure you, we get *way* fewer than 1600 PhD applicants interested in scientific computing (especially in CS -- a lot of those who are interested in my type of work go toward applied math).
Prikaži ovu nit -
And there is a lot more to the world than CS, too. Yeah, computers are great, the work has impact, etc. But look up from the screen, dammit! There is *so much* stuff that is fascinating in the world, and important, and needs research.
Prikaži ovu nit -
We need people studying social sciences and humanities. Look around the net -- do you really think over-emphasis on ML is helping us make the world we want to live in?
Prikaži ovu nit -
There are fascinating hard problems across engineering. Build a better battery! It's way harder than building a better DNN. Figure out cheap, sustainable building materials without the greenhouse implications that concrete has. Figure out the building blocks of biology.
Prikaži ovu nit -
Fifth: Academia is *big*. I'm never sure what to make about general complaints about academia, which seem to me to be roughly as useful as general complaints about business as a whole. Different departments and institutions are very different.
Prikaži ovu nit -
I know what is true for Cornell CS, for a handful of other places where I have direct experience (in CS, math, engineering, and statistics), and for a larger set of places where I have friends I talk to (across a broader set of fields). But that leaves out a ton.
Prikaži ovu nit -
Even in one field and sub-area (CS ML), the idea that all schools have identical tastes and processes in PhD admissions -- and that these involve paper counting -- is... incorrect, I think.
Prikaži ovu nit
Kraj razgovora
Novi razgovor -
Čini se da učitavanje traje već neko vrijeme.
Twitter je možda preopterećen ili ima kratkotrajnih poteškoća u radu. Pokušajte ponovno ili potražite dodatne informacije u odjeljku Status Twittera.