Hacker News Reader

Comment by intenex | Hacker News Reader

Fairly shocking to me that Yuan Cao received literally a single glancing mention here, but he was clearly the one who actually made the pivotal discovery, and is the first author on the Nature paper (https://www.nature.com/articles/nature26154), and was the one named to the Nature 10 in 2018 for his discovery.

Really mind boggling that this entire piece focuses all the attention on his supervisor and almost neglects to mention him at all.

View on HN

Usually the grad student does most of the work, but the ideas and direction comes from the supervisor. It's normal that the supervisor funds, directs and communicates the work that's being done. More often than not, the supervisor has a better idea of the big picture and the implications of the work than the grad student, so it's natural that the supervisor talks to the press.

For example, I'm a grad student, in a department where my supervisor would talk to the press on work that I would have done.

In CS related fields, in particular ML and CV, this is absolutely not the usual situation except perhaps in the beginning of the PhD. Perhaps it's more common in fields where the research is done on top of a heavy investment in infrastructure/equipment that is lead by the professor, e.g. in physics and biology?

In my experience in biology new professors provide a lot of ideas, old professors just steal new ideas from their students and post docs.

> Yuan Cao ... was clearly the one who actually made the pivotal discovery

Are you saying that based on something external to the paper? In the author contribution section, Cao is not credited with having done anything on his own:

> Author Contributions

> Y.C., J.Y.L., J.D.S-Y fabricated the devices and performed transport measurements. Y.C., V.F. performed data analysis. P.J.H. supervised the project. S.F. and E.K. provided numerical calculations. S.L.T., A.D. and R.C.A. measured capacitance data. K.W. and T.T. provided h-BN devices. Y.C., V.F., and P.J.H. wrote the paper with input from all authors.

The first authorship is an important meaningful thing, but it definitely isn't a guarantee that they key, most noteworthy step was done by the first author, much less the first author alone.

Pretty typical of academia. Grad students also don't normally get Nobels

And for good reason. A responsible supervisor will typically give PhD students tasks that he has already "kind-of / sort-of" worked out as feasible. I once read the following recommendation for how to set dissertation subjects, I can't recall by whom.

- BSc thesis: what I can do over lunch

- Master thesis: what I could do in an afternoon

- PhD thesis: what I could do in a week

(Here the figures don't mean typing, programming etc, but the core intellectual work). After having supervised nearly 200 theses are BSc, Masters and PhD level, I must agree, this is a pretty great heuristic.

Note: there are exceptions. Some PhD theses go way beyond this, but those are rare. I have no idea about Yuan Cao's work, discussed here.

Where are you in the world that the level of sophistication of the work being set is such that this heuristic holds true?

I suppose a BSc thesis or design report etc. is supposed to be more of an exercise in demonstrating some understanding and creating something, but beyond that I would expect some fairly novel impact from an MSc and definitely from a PhD.

I'm very surprised by this as the quality of MSc theses and PhD theses at the universities I attended were all fairly novel, and even if the supervisor had suspected the same conclusion, the amount of work to arrive at that conclusion is non-trivial when doing research.

I fear this devalues the contribution of what these young researchers are doing. I know many people who went on to very good research positions and even founded companies based on their Masters or PhD work.

None of this applicable for hard experimental sciences like physics and chemistry. You can't shortcut bench time, especially since it involves a lot of trial and error. And sometimes it doesn't work out, at all.

What field are you in? This applicability of this strikes me as wildly different depending on the field and the standards of the university.

Cynical me asks isn't that how it works for PhD students?

Yes, probably.

Reply on news.ycombinator.com