Hacker News Reader

Comment by dfgert | Hacker News Reader

I disagree a bit with you. In today's world we have reached a state where next new research output requires access to advance tools and strong funding which is only available to researchers at flagship institutions. There isn't any real research happening outside of flagship institutions; other institutions may make big claims on research they are doing but they almost never deliver output. In some sense, distribution of nobel prizes per country can be attributed countries research outout.

View on HN

> There isn't any real research happening outside of flagship institutions

Does that really imply research only happens at stanford, MIT and Harvard? Or what are "flagship" institutions?

What I wanted to say is that you can skew you distribution of researchers towards the absolute top by giving the few an enviroment the competitors can't match. I would think that this will probably be revealed if you compare for example nobel-prizes to overall citations a country produced. But that's harder to track.

>Does that really imply research only happens at stanford, MIT and Harvard?

The research that matters, yes. There and in a few others all around the world.

Papers, on the other hand, are written everywhere by the thousands...

Well, then I have to vehemently disagree with you. The bulk of interesting papers I read are not from MIT or Harvard, but from a quite diverse group of interesting universities and non-university research institutions, from Canada to France to Switzerland. There are a lot of places where real and important research is done.

I would even say the opposite. I think a lot of papers from the top-institutions are quite boring and, i think, heavily profit from the name of the institution.

There are quite a few countries with many strong universities, for example, the UK, where I regularly enouncter universities (I think) I have never heard of before. Right now, I have a paper open from the University of Strathclyde, I can't prononounce the name and it's definitly not MIT, but it's definitly research that matters.

> The research that matters, yes. There and in a few others all around the world.

Do you have any basis for this claim?

I've found that (for example) the top research in cosmology and the top in material design and the top in semiconductor physics (as it moves into exploring exotics) are not only different universities (around the world), but that none of them are the three that are mentioned up-thread.

Now, I can sling an anecdote with the best of them, but I am surprised at your claim, so I'm hoping you have better data than I?

>but that none of them are the three that are mentioned up-thread.

Not sure if cosmology is some exception (I guess one could find specialist universities in all kinds of subjects), but overall those three as exactly ranked as the top three, ranked by papers, quality, citations, patents, etc, here, across all kinds of fields:

"To compile the 2017 ranking of the world’s most innovative universities, Clarivate Analytics (formerly the Intellectual Property & Science business of Thomson Reuters) began by identifying more than 600 global organizations – including educational institutions, nonprofit charities and government-funded labs – that publish the most academic research. Then they evaluated each candidate on 10 different metrics, focusing on academic papers (which indicate basic research) and patent filings (which point to an institution’s ability to apply research and commercialize its discoveries), and ranked them based on their performance."

1 Stanford University

2 Massachusetts Institute of Technology (MIT)

3 Harvard University

https://www.reuters.com/article/us-amers-reuters-ranking-inn...

this is exactly what I wanted to criticise by my comment. I don't think these rankings are that important because they don't capture what the (original) article wanted to measure. There's only one stanford, which might be better than every other university out there. But there are a lot, often way smaller, universities out there that produce fantastic research. You have to add them together.

I'm curious, in your experience, (and I am curious - not being argumentative), how socially connected do you find the citation graph to be?

A while back, I was mapping innovation diffusion, as indicated by papers and patents and hiring company's R&D budgets and ROI on said investment and the like, and I was surprised at how much innovation spreads through folks that know one another, as opposed to papers that anyone could (in theory) read. I have to guess that people would similarly cite papers of people they know (maybe even of the same lab?). Do you think that has any effect on the score that you are pointing me towards?

Oh come on, everybody can churn out papers and patent filings.

If you go by Nobel prizes, the ranking is different (leaving out non-US contestants):

Physics:

1. MIT

2. Harvard

3. Berkeley

4. Chicago

5. Columbia

Chemistry:

1. Harvard

2. Berkeley

3. Chicago

4. Caltech

5. MIT

Medicine:

1. Harvard

2. Rockefeller

3. Columbia

4. Caltech

5. WashU

Conclusion: Stanford is overrated.

This, there are experiments that cost millions of dollars just to build and they can fail and need many other costly experiments. Beyond this, it is true that some researchers can find theoretical research using only "pen and pencil... and computers".

OñAnother example about the former case (in computer science) is where a researcher can quickly build a team of a hundred post doc students to help in specific topics.

I don't think there are many (any) academic labs in CS with 100 post-docs.

also post docs arent students.

big engineering research efforts like that DO happen in industry though.

There's a big difference between how much it costs to do research in CS and how much it costs to do research elsewhere. It costs $10^9/year to run the LHC.

the comment I am responding to is talking about computer science

Expensive experiments are needed only in physics and medicine, unless I have wrong notions about chemistry, out of the six Nobel prize categories.

Reply on news.ycombinator.com