Syllabus for Eric's PhD Students

As a rising 5th year PhD in ML -- I could not agree with this advice more! I have very hands-off advisors. I spent the first 3 years "wandering the woods to find something". Last year, I really had to sit down and think about how I can finish up my PhD on time. I pretty much did what you outline here. A lot of tools I used were organization tools I learned from my business school/product manager friends.

I honestly think the PhD system needs to be overhauled. even at a "top-tier" program like mine, it is amazing to me that at no point do we receive any training regarding practical components of research. 100 years ago, the way you became a physician was to follow around a physician and one day you were ready to be a physician yourself. In the year 2020, this is how PhDs are trained. I do realize that a PhD is not a "professional" degree like MD or JD, however, given that most PhD grads will 1) go to industry 2) go into academia, we need to teach students about project management, planning etc.

Just like mine :)

There is a link on the doc to that website ;)

While I agree with his perspective on classes, some departments are hostile to this.

When I was in grad school, I was required to take a ridiculous number of course credits to satisfy the requirements. Moreover, if you wanted to pass the qualifying exam, you had to know the material in some of those courses really well. The professors always complained that the requirements were too high and many of their students would lose 2-3 years just satisfying them. But ... professors are the ones who made those requirements to begin with!

> Concretely, even if you don't know what the answer is going to be at the end of your research, you must think about, or have an idea about, the format of what that amazing answer is going to be. Write the outline of your thesis and "ghost out" what the major charts will be. Write the intro sentences of each chapter -- what are they? (and I don't just mean the boring review of the field part, but your findings part)

> You should know what major type of finding, plot, or table your research is going to output. What are the columns and rows of that table, or axes of that plot? How many data points are required? How many of them can already be guessed? Where is the surprise going to be? What is the conclusion going to be?

You are absolutely right and it's depressing.

The only practical, predictable and steady way to a PhD is to come up with the storyline in advance, do the experiments, bend the interpretation and analysis to support the story you already wanted in advance, then pretend the finding is surprising and novel (when actually you knew you could safely make it work in advance).

There is a tension here, on the one hand PhDs have to pretend play that they discovered major novel stuff and innovated a lot (which is only possible through immense risk), on the other hand in reality they must finish in a given number of years and must be very risk averse.

Only a major cultural shift towards handing out PhD titles for meticulous, systematic, risky but unfruitful research may help. Forcing people to find flashy new stuff will lead to tons of overinflated findings that are more about storytelling than actual surprising novel things that bring a field forwards. It's all a huge pretend play.

----

Ideally in an alternate universe of different incentive structures that support real science, a person would work in research, use existing approaches in different scenarios, push and prod them in different use cases then perhaps once every year you'd stumble upon some interesting effect you don't understand. Then you'd dive in and prod this specific effect with well designed experiments. Once it's clearer what is going on, you come up with the best way to convey this newly earned knowledge. It is at this point (once you actually know what you want to alert the research community to) that you decide how to best present the knowledge, what tables, sections and paragraphs will succinctly and ergonomically convey the information.

If you knew all the story and layout beforehand and you know you will be able to just plug in the numbers and make the paper stand, then you are not doing real science, you are playing the academic game, a cargo cult of science, a game stealing and consuming the prestige of science.

Yes, as I said this only works in the alternate universe. Here in this universe we must churn out papers and must overinflate our findings. Of course, we need to eat, we need jobs, we need titles etc.

It has driven so many PhD students I know to bitterness and cynicism. Several have given up and went to industry. Others, after the disillusionment, have cynically pushed through with whatever minimal things they knew would be required (plus the right story and presentation) then bailed to consulting. You also need to bullshit in consulting but are at least compensated for it well enough.

In the rarest cases do people come out of it as a positive passionate person. Even the successful ones become a hedging, scheming, lukewarm, slimy academic personality. Everything encourages this. The way you get professor positions, the way you get grant money etc. It optimizes and chisels people into becoming the antithesis of curious, open scientific inquiry.