Apparently, prestige matters in CS

[thissiteispoison]

If you want an academic job, that is: http://advances.sciencemag.org/content/1/1/e1400005

 

Something to consider while making decisions, unfortunately. 

[fuzzylogician]

In other news, the sun is hot.

[TakeruK]

First, I don't think it's necessarily a bad thing that prestige matters. Sure, I agree that hiring based on prestige reflects social inequality and that academia is not at the ideal quality level. But, other workplaces in the world are also not at equality. I don't think there is any reason to expect academia to be special and different from the rest of the working world. 

 

Second, I do not believe a meritocracy is the ideal state for academia or any working world. I am in favour of a system where resources/materials are granted to those who need it, and I think that in many cases, the distribution of resources that is the most fair/equal is not necessarily the same distribution as the most meritorious. I think building a system based on merit is not ideal especially since many studies show that humans are not impartial and implicit biases impair our ability to determine merit accurately. 

 

And finally, I do not agree with the authors' statement that professors at higher ranked schools should produce more papers/research etc in proportion to placement rate (see bolded statement from quote below). Why should it be a linear relationship between scholarship production and professorship production? I would think that a logarithmic relationship makes more sense. If scholarship ability is normally distributed amongst the population, then I don't think this linear relationship makes sense. And we don't really have a very good way of measuring "scholarship output".

 

Strong inequality holds even among the top faculty producers: the top 10 units produce 1.6 to 3.0 times more faculty than the second 10, and 2.3 to 5.6 times more than the third 10. For such differences to reflect purely meritocratic outcomes, that is, utilitarian optimality of total scholarship (13), differences in placement rates must reflect inherent differences in the production of scholarship. Under a meritocracy, the observed placement rates would imply that faculty with doctorates from the top 10 units are inherently two to six times more productive than faculty with doctorates from the third 10 units. The magnitude of these differences makes a pure meritocracy seem implausible, suggesting the influence of nonmeritocratic factors like social status.

 

 

Overall, I agree with the authors that the faculty at the top schools are not super-beings compared to faculty at lower ranked schools. I am at a top school and the administration here openly admits this. They say that our faculty are the best not because they are special in some way, but because the school provides with them all the resources necessary to succeed and produce great science. A meritocratic perspective would be to ensure the people my school hires are actually the very best so that they make the best use of the resources. However, I think an equally valid perspective would be that we just need to hire people that are "good enough" to not waste the resources. In addition, since there are other aspects to being a researcher than scientific production (such as teaching, mentorship, being able to work well with others), I would say it's more important to consider other factors besides merit.

[thissiteispoison]

CS in particular tries to pride itself in being a meritocracy, and when I started applications everyone and his mom was like "rankings mean nothing, it's all about advisor fit." Rankings, though not precisely the same as prestige, clearly do mean something if you want an academic job. Sounds like I'm stating the obvious, but really, computer science as a field really likes to pretend it's above subjectivity. 

[phdreally]

@thissiteispoison - I could only agree. There clearly seems to be a contradiction. Everybody on this forum say "ranking means nothing, it's all about advisor fit." And then, "ranking matters only if you go to the industry". And now the paper you cite says "ranking matters if you want an academic job". If not industry or academy, what else does one keep in mind while deciding and ignoring the ranking? Maybe sheer "intellectual satisfaction"?

[thissiteispoison]

Given your list, you'll be fine with any choice. Take a look at the supplementary materials, figure S10: http://advances.sciencemag.org/cgi/ content/full/1/1/e1400005/DC1

 

It gives the schools a prestige score. Possibly more interesting than raw rankings. 

 

Figure 1 (flow for the top institutions) is also pretty neat. The nerd in me (all of me I guess) loves this entire paper as depressing as I find the reality it reflects. 

Edited March 10, 2015 by thissiteispoison

[Loser777]

Interesting how the relative positions of Caltech and UW are flipped on this list...

[pascal_barbots_wager]

Do I have this right: the paper is saying that, according to some measure of scholarly productivity, graduates of the top 10 schools are not so much more productive than graduates of the top 10-20 or 20-30 etc. that their higher representation among faculty can be explained by productivity alone?

 

Ignoring the validity of that measure (which as TakeruK mentioned is a pretty big question), isn't this kind of a misleading question? The number of times you choose X over Y doesn't really correlate with the degree to which you prefer X over Y, or how much better X is than Y. X might be just slightly better than Y, and that might be enough to get chosen over and over again. 

Edited March 10, 2015 by pascal_barbots_wager

[Icydubloon]

It doesn't matter if it is just slightly better or 1000x times better. The max is the max and will be selected each time.

[vertices]

The paper doesn't measure productivity at all. It just says that the productivity implied is unrealistic. As others have pointed out, the way they view this is up for debate.

 

There are many other problems. One, this is a small numbers thing and is likely easily skewed especially as you go down the list. Two, this is measuring things that have occurred infrequently (small numbers) over a very long period of time. One school could have done great 10-30 years ago but not so well with their recent graduates.

 

As for "advisor fit," I still believe that's first and foremost. If you go to a prestigious school with poor advisor fit, you will suffer for it -- however, your labmates will be much more likely to become professors! The key is to find good advisor fit with an advisor that has good connections and good resources. At prestigious schools, most advisor will have good connections and resources. However, you can find advisors like that at not as prestigious schools. Sadly the numbers are probably way too small to do an advisor ranking this way. I would love to see it though.