Hi Phil,

I think that it is appropriate for a graduate student to use a state-of-the-art tool in their research effort, without having to build that tool themselves. The keys are that they understand the tool very well (know when it is likely to fail and why) and that they have complete "ownership" of the research that they are doing with the tool. By the time the work is completed, the student must be one of the worlds experts in that question. They have spent more time thinking about than anyone else, and that had better be obvious in the papers and presentations that arise from the thesis.

The question of whether the research is relevant is something both the student and the advisor should consider in the early days of the project. I think it is challenging for a student to comprehend the implications (or irrelevance) of a complex problem.

"Should we be driven mostly by tools at hand and by societal needs? Do curiosity and the strictures of The Scientific Method have a role to play?" Well its really hard to do detailed work on something you find to be boring, so if you aren't curious about it your unlikely to do a credible job. It is also easy to identify interesting problems that can't be solved, having a state-of-the-art tool available is often required for progress to be made.

Most of us work on projects that are externally funded, but that does not mean is is "driven by those outside of science", rather it is a process of convincing your colleagues that the project is interesting, relevant and doable.

Finally - as a community we need to appreciate those who spend substantial parts of their careers building the state-of-the-art tools. Instruments, codes, methodologies and algorithms are often produced by specialist. Commitment to producing these tools often results in a lower publication rate, something that members of review committees should keep in mind.

Ed

Thanks for diving in Ed! I appreciate the thoughts

very much.

> I think that it is appropriate for a graduate

> student to use a state-of-the-art tool in their

> research effort, without having to build that

> tool themselves.

this was not really the question- I was asking if

we should be *driven* by the tools we have, in

selecting thesis problems, for example, not if it

is appropriate to use one or not.

To be clearer, I worry that some research is like

the carpenter who has a hammer, so all he can do

is hit nails... In the same way I worry that too

little time is spent identifying really meaningful

hypotheses in space physics that can at least be

falsified.

> I think it is challenging for a student to

> comprehend the implications (or irrelevance) of

> a complex problem.

Yes yes, well said. But is this not the essence

of doctoral research work over and above other

endeavors? Is a doctorate merely an

apprenticeship for people to use how to use tools?

Or is its purpose to challenge authorities

(advisors etc) to go beyond what they thought

possible? Both are important, but I see the

latter as progressively getting less

important. This I find worrisome.

> Well its really hard to do detailed work on

> something you find to be boring, so if you

> aren't curious about it your unlikely to do a

> credible job.

Unfortunately I know students who are seduced by

playing with computers and data, the subject (the

Sun) is sometimes of marginal interest. If you

like, it is the "video gaming" aspect that is so

attractive to many people. This must be a

dangerous thing for the future of solar physics.

Also, others are attracted by being able to

predict solar behavior. But we cannot say this is

solar physics given that we cannot even explain

why the sun forms something like a spot at all.

Don't get me wrong, there are some fascinating

advances in computer algorithms and predictive

tools, but let's not pretend we're getting to the

bottom of some real problems in solar physics.

Was the last real breakthrough in solar physics

the determination of the internal rotation

profile of the Sun some 20-25 years ago?

> It is also easy to identify interesting problems

> that can't be solved, having a state-of-the-art

> tool available is often required for progress to

> be made.

Ah, I agree, except that the "state of the art

tool" we must use *first* is that between the

ears.

> Most of us work on projects that are externally

> funded, but that does not mean is is "driven by

> those outside of science", rather it is a

> process of convincing your colleagues that the

> project is interesting, relevant and doable.

Good point, but I believe that the pressure for

society to have solar people make predictions is

precisely this. The question is, as physicists

looking at this problem, what do we *do* about

this pressure? Given the state of solar

predictions, I do not think this endeavor should

count yet under the realm of "science."

> Finally - as a community we need to appreciate

> those who spend ... committees should keep in

> mind.

absolutely, well said.

I wonder if anyone else follows this stuff..

Dear Philip, dear Ed,

Thanks for the nice discussion.

I have the feeling that what Philip asks is sort of a tendency in many disciplines, at least in Earth Sciences. Being an interdisciplinary scientist (if I may) I observe the same tendency almost everywhere. The reason is mostly the projectism and partly the bureaucratic burden of the supervisors. The supervisors should offer a "bulletproof" topic *that can be completed* in the given time, no risks as much as possible, the "tools" are there (happily funded by the project), dear student, just provide the results, report, publication, done. (Check the fantastic comic strip of Jorge Cham at phdcomics.com, e.g. a recent one, comicid=1636)

I think most of the supervisors would enjoy thoughtful discussions with their students, but only very few of them can find time for that. (The latter may partly apply to the students themselves.) I remember when my supervisor discussed my new ideas with me (well, ages ago), what a revelation it was.

I must admit that I myself have problems to find time for discussions with my students. Fortunately, most of them are aware of the danger of "pushing the button" and "the computer gave the following results..." Hollywood-like behaviour.

> Ah, I agree, except that the "state of the art

> tool" we must use *first* is that between the

> ears.

That's really great, thank you. I will cite that to my students, with citation reference, of course.

I try to attract those students who are interested to know (a) what is in the box (b) they want to build such a box, or even (c) build an improved one. These students really use their tool between the ears. After all, this is our goal, isn't it?

Balázs

Every student is different of course (and structure and length of PhD is different in different countries), but with my students recently I try to start them off for first year on a mini-project looking at an event or events, probably using other people's software. By end of mini-project you know the student's capabilities and weaknesses whilst they have learnt a lot on the field and other skills. I think it's pretty important that the student has a chance to develop their own code later in the PhD. Surely one of the most powerful aspects of research today is being able to solve a problem by directly writing your own code.