T-Shaped People and Academia
bastian.rieck.me
bastian.rieck.me
I actively advise students against joining labs like this and consider the whole enterprise unethical. I raise concerns whenever I'm reviewing grant proposals by folks who run their labs like this. I've successful nuked funding proposals because the work plan included language like this.
Want a devops engineer? Hire a devops engineer ("systems scientist" or so on). Want a graphic designer? Hire a graphic designer.
But DO NOT lie to impressionable youth by telling them that you are admitting them into a PhD program to be your apprentice on a research project and then proceed to pay them $30K/yr (with marginal benefits) for uninteresting and unexceptional engineering and design work. They are taking a huge financial opportunity cost to work on uniquely high leverage problems. Maintaining your research group's build system is not what they signed up for.
> ‘I’m a researcher, not a(n) X,’ with X∈ {software developer, designer, marketing person, artist, …}
Sure, but here's the rub. What does everything in that set have in common? They make at least 2x+ the average PhD student and have much better benefits.
If you want a "force multiplier" person in your lab, do your job: get a grant and pay a fair price for that sort of labor. PhD students are NOT ∈ {software developer, designer, marketing person, artist, …}. If they were, you'd be paying quite a bit more and/or getting far less exceptional candidates.
My observation is that modern machine learning research entails so many additional moving parts (most, but not all of them, involving around SWE) that any student with skills in this area is already an asset. However, these skills are typically ignored while hiring.
I have seen the scenario you describe, but I it is definitely not what I intended to put a finger on in this post. Thanks for pointing this out!
Then you must tell them that when they do “non-core-work“ they are essentially wasting their time; at least if they intend to stay in academia.
Sure there is some technical housekeeping necessary to get research done and papers written. But it must be kept to an absolute minimum.
It is nice to have students who polish the web site or set up CI on GitHub. But these students will almost certainly have fewer research results to show at the end of their PhD, compared to their officemates who just focus 100% on their research.
I don’t like it either but it’s the painful truth. I’ve seen way too many PhD students wasting their time doing incredibly valuable work for the lab but failing to publish their own work. It‘s the responsibility of the PI to make it absolutely clear to those students what is happening and that they should focus on a career outside academia if they enjoy doing such things.
However, I find it somehow weird that we only ever measure success along one axis. Moreover, these students often serve as facilitators, making it easier for others to finish projects. Navigating this precarious space is not an easy task for me, I admit that.
But honestly, your options are very limited, especially in Germany. You cannot pay them more, and you also cannot offer them a permanent employment contract. Most of the time it’s even hard to extend an existing contract. You can perhaps send them to nice conferences, but then you would send the principal author to the conference, too.
But I totally agree that academia fails at rewarding people who facilitate the work of others. There is some perversion to this, since the key part of academia is precisely to spread and exchange knowledge.
In this light it is funny to see that in many universities (especially in Germany) teaching is not valued much for career progression.
What you’re describing is glue work[1].
Until recently, glue work tended to be invisible and unappreciated across a good majority of the industry. While it’s starting to get visibility, there still isn’t consensus on how to recognize it and/or reward it.
> But I totally agree that academia fails at rewarding people who facilitate the work of others. There is some perversion to this, since the key part of academia is precisely to spread and exchange knowledge.
Yes. I saw quite a few talented people turn the backs on a career in academia for exactly this reason.
> In this light it is funny to see that in many universities (especially in Germany) teaching is not valued much for career progression.
Also yes---but that part is at least changing a little bit (even though the predominant thinking is still 'Yeah, who cares?') in the younger generation of PIs.
Things are changing extremely slowly in academia. It’s always the old faculty that decides who becomes new faculty, and their decisions tend to be very conservative. What’s worse, I have seen very progressive young PIs become complacent and conservative once they had tenure. I wouldn’t bet on academia changing its tenure criteria.
That's what I am worried about in myself, having seen this in others as well. I'll try to be more introspective.
> I wouldn’t bet on academia changing its tenure criteria.
No, I am not doing that either, but in this case, there's always the 'War Games' games response: 'A strange game. The only winning move is not to play.'
And this may depend where in Europe, but they get paid more than in the US, and also tend to get full benefits.
At least those that I knew in Denmark/Belgium were. They were not even classified as students, but as full time employees. Doing PhD research was their job.
On the one hand, you recognize in your article that "non-core" work is not rewarded in academia. In particular, however, the situation is dire for current PhD students. Getting into a PhD program is easy compared to getting an academic job (even - or especially - in the EU) with at least some job security. Further, you probably know that most PhDs lack the knowledge, experience and ability to accurately judge their chances on the future job market with any accuracy.
On the other hand, your replies indicate that you are following the "traditional" European research hierarchy, where the PhD student does years of "non-core" work for the Professor, ultimately ends up with less publications. This leads to issues competing internationally, a chain of short term contracts, post-docs or adjunct positions and finally, many leaving academia (or at least the EU). Which, by the way, is a real problem.
In European academia, we are all aware of these traditional hierarchical sentiments. At each university, there are stories about the Professor who said something like: "Your first research ideas belong to me", or Professors who actively resist the establishment of a stipend-based PhD program with the argument that "Someone (the PhDs) has to do the work at the institute!".
Now, there are probably countries that do better than others. Your state might mandate that PhD students get a salary and benefits, but all that is worth little if they end up in a bad condition after 3-6 years of doing their PhD.
And here, your reply here worries me. No, you do not hire PhDs to *work* for you. You hire them to train them to be researchers and to collaborate on that. Next to publishing and teaching, as a Professor, you are responsible for enabling these PhD students to have a successful career. More than in any other career, it is YOU, the advisor, who ends up making or breaking the future of a human being.
The thought that PhDs are merely hired workers who have to work through "menial, non core" work before they are allowed to get to their research, is one of the main reasons why academic research in the EU is doing so badly compared to other regions in the world.
In the interest of your future PhD students, I implore you to consider how much impact you really have on their lives!
I am not sure where you are getting this from. My intent is only to point out that there is more to consider than publication output. I will try to make this position more precise.
I again want to stress at this point that I do not want to hire students to perform non-core work; this is not what I wrote in the article and I am certainly not advocating for that because it is exploitation (almost of the form 'trading a PhD for years of programming'). My point is that modern research has some non-core aspects, and in my experience, these are ignore by the PIs insofar as they are expected to be performed, but they will not be rewarded or recognised.
> On the other hand, your replies indicate that you are following the "traditional" European research hierarchy, where the PhD student does years of "non-core" work for the Professor, ultimately ends up with less publications. This leads to issues competing internationally, a chain of short term contracts, post-docs or adjunct positions and finally, many leaving academia (or at least the EU). Which, by the way, is a real problem.
Fully agree with you that it's a problem. But in my experience, we cannot get around the fact that ML research is certainly helped by good SWE skills. Now there are people that like these sort of aspects more than others; and my point is that they can also help to enrich a team, even though---if just seen through the lens of publication output---they might seem less strong.
> In European academia, we are all aware of these traditional hierarchical sentiments. At each university, there are stories about the Professor who said something like: "Your first research ideas belong to me", or Professors who actively resist the establishment of a stipend-based PhD program with the argument that "Someone (the PhDs) has to do the work at the institute!".
Agreed, but I have to emphasise that I am not talk about doing work for the institute or for maintaining lab infrastructure. I realise how one could arrive at this impression, but this is far from my original intent. I wanted to point out that hiring someone with a more generalist skill set and mindset can enrich a research time _just as well_ as hiring someone with a more expert mindset.
> And here, your reply here worries me. No, you do not hire PhDs to work for you. You hire them to train them to be researchers and to collaborate on that. Next to publishing and teaching, as a Professor, you are responsible for enabling these PhD students to have a successful career. More than in any other career, it is YOU, the advisor, who ends up making or breaking the future of a human being.
If I could rephrase this in my own terms, I would also state that I hire them to work with me, to collaborate with me. I also agree that I should enable them to have a successful career---regardless of where they choose to work in the end.
> The thought that PhDs are merely hired workers who have to work through "menial, non core" work before they are allowed to get to their research, is one of the main reasons why academic research in the EU is doing so badly compared to other regions in the world.
Fully agree with you here, but this is not what I wrote at all. A PhD student is foremost a budding researcher. My point is that it makes sense to consider their skill sets in a more holistic fashion when hiring them.
> In the interest of your future PhD students, I implore you to consider how much impact you really have on their lives!
I am keenly aware of this responsibility. I realise that my points above might have been misconstrued, but please understand that I have foremost their interests in mind. It is disheartening to see how my more generalist friends who are doing their PhDs are not recognised at all in academia. _This_, IMO, is also some form of 'brain drain' that could be avoided.
(And yes, I spent time on logos, websites, and whatnot that never make it into papers: it all adds up and you never really know what is ok to drop until after the fact. More so, the paper is the worst form of result and better as just a stepping stone, IMO: a paper is the failure to do anything more meaningful & expansive, like a community of practice or company.)
> reviewers assume they know best.
You are probably more adept at running a grant-funded research group than me. I have no doubt. My concern here is ethical. I do think the average PI is in waaaay too deep to admit how horrible even the "good" PIs end up treating their phd students. So, in that sense, I do generally assume I know best. Not about how to do successful research. But about the true reality of how the people in those salt mines are being treated.
On that note: of course all that non-core work is valuable! Incredibly valuable! And PIs should probably hire someone to do that work! At fair market wages! With accurate job titles and job descriptions!
I agree with everything you're saying... except about who should do the labor and what they should be paid for that labor.
I'm going to reiterate two sentences from my original post.
>> Want a devops engineer? Hire a devops engineer ("systems scientist" or so on). Want a graphic designer? Hire a graphic designer.
>>> ‘I’m a researcher, not a(n) X,’ with X∈ {software developer, designer, marketing person, artist, …}
>> Sure, but here's the rub. What does everything in that set have in common? They make at least 2x+ the average PhD student and have much better benefits.
>> If you want a "force multiplier" person in your lab, do your job: get a grant and pay a fair price for that sort of labor. PhD students are NOT ∈ {software developer, designer, marketing person, artist, …}. If they were, you'd be paying quite a bit more and/or getting far less exceptional candidates.
Labs that have successful spin-offs are usually the worst about this sort of thing. They're running their research lab like a seedling incubator, but paying each of the one man shops printing their lottery tickets low five figures without retirement benefits.
If you want to go ethical -
* PhD students, most of all, are learning how to do R&D: picking valuable problems, picking how to solve them, and most efficiently in time/money/etc solving then. Doing scrappy Day 1 work is very much a part of that skillset, surprisingly similar to a founding MVP team: Not My Job is teaching the wrong mindset and divorces you from the community + work, which may be what you are trying to learn.
* ... However, Learning to Outsource/delegate is indeed part of efficiency training, so let's dig in. The default answer in most prototyping is do a crappy cheapo version, replace with a more costly version when it becomes the need, and reserve your shoestring budget for a few really important things. So likely no to a commercial grade design firm on rerainer, yes to either a $50 logo on Fiverr or more likely, something funny you can make via a free tool, done. Later, when there is more need and budget, say Year 2, easy to revisit.
* Some things do merit staff. Should a bio lab partner with a CS team who wants to learn their needs as part of making new CS methods? That's T-shaped collaboration and good, and part of why academic teams can be so scrappy: there's ~always someone somewhere where your non-core work is their baby. Or, maybe they want to inhouse to get more speed/reliability/control. If that person will be like the CS team, ultimately making new CS methods, or a more computationally minded bio researcher helping with some of their own coding for faster EDA, great. But if they will always be Author #2 and has to code on projects they are not doing the science on, that's not a postdoc, that's a staff programmer. We should have more of those.
Ex: My wife's lab had paid staff, but she still had to feed cell cultures at 1am for the research she wanted to do, and later decided those methods were not for her and no longer does wetlab projects. That doesn't invalidate people still doing cell cultures as part of their methods, and I'd still love to see more robots & no-code & budget there to make it easier.
Edit: RE:products,
Whoosh. Academically derived products at that scale imply a community of practice and development: societal impact and dedicated makers growing it, including more knowledge. That should have the level of respect that a paper that opens up a small subfield gets. From a norms perspective, it's disappointing to see what sounds like a practicing academic to not value that. Separately, again, community of practice work values being embedded, so a student not enjoying it is their personal methods choice, not rules for others.
You're promoting both a super traditional view yet also one that seems divorced from the last ~100 years of sociology studies wrt how society progresses wrt inventions and how people fit into that. So, it's great you follow old beliefs handed down by generations of giants from much earlier ages -- there's clearly precedent -- and I encourage you to learn about new ones from the last few generation of scientists who took a more scientific method to learning how recent progress often works. These methods are not for everyone, but it's useful to at least know about them before calling the techniques of your peers 'cancer'. Personally, as technology keeps speeding up, I expect this to become even more important, but at the same time, do value head-in-the-sand research too ;-)
Also, to be clear: I am not rejecting the scientific method here. Products, and more importantly, the communities of practice around them, have a lot to do with hypothesis generation, methods, and evaluation. Most industrial labs, and ironically, especially ones that force researchers to embed into existing internal product teams, put a lot of hand cuffs on this process so academic env helps. It does take humility and empathy and time to incorporate them, so not so easy (kind of like startup-lite), but it is a family of tools and one aimed at more intentional progress.
And no, 'products' is not synonymous with financial motivation: most OSS projects have no paid piece, for example.
PS: And upon writing this comment, I had the epiphany that 'pushing out papers' might indeed be the primary objective for some. :-|
Yes, of course they do. The whole point is that this type of labor is valuable.
> if your goal is to build something that people actually want to use, as opposed to just pushing out papers.
Neither of those things is the purpose of a phd...
If a student's goal is to build something that people actually want to use, they are much better served by taking an industry position in a product group.
> pushing out papers
From my perspective, the obsession with publishing is just another version of the same problem.
Startups are kind of like that, but still quite different because VC/P&L/growth/sales still warp things. (And PhD programs are great places to experiment before that jump, as the learning & investigation can take an unknown amount of time.)
Agreed, I should have phrased that better. Fundamentally, the focus should be on 'generating new knowledge,' right?
> From my perspective, the obsession with publishing is just another version of the same problem.
Yes. :(
Some biology students and postdocs have described to me an antagonistic relationship to the funder or PI goals where they dismiss their responsibilities for data quality and reuse. For them, it isn't on the critical path to their paper and their career. Or, having data in disarray is, in their minds, giving them an edge to not be "scooped" in the lab. These cynical attitudes are almost subversive, in how the junior worker tries to evade or sabotage aspects of their job responsibility which do not serve them personally. I also see some who seem to engage in magical thinking, as if some other staff or informatician should come by later and clean up their mess of data, without the researcher putting in the effort. I often wonder whether they really misunderstand the problem so severely, or merely use this fig leaf to avoid feelings of shame.
I know that lines should be drawn between different roles, but there are inappropriate expectations on both ends of a spectrum. PIs should not treat junior researchers like assistants to get coffee, run personal errands, etc. But neither should junior researchers believe that the day to day grind of the scientific process is beneath them. At the scales that most labs work, there are many essential yet mundane tasks that need to be shouldered by "core" participants. The overhead of trying to delegate these tasks to a different person would either require more time than doing the task oneself, or result in abysmal quality when the hand-off is made with insufficient supporting effort.
I think that people exposed to the pop culture of startups and large corporations can absorb a mistaken idea of how easily human organizations scale. With vast numbers of skilled staff, you can create processes which scale beyond the capabilities of a typical small lab or group. These tend to develop lots of specialized roles to support that process. But, I think there is a huge inflection point between the efficient small group and the scalable large organization, where the marginal value of adding another staff member can be very small, or even negative. Many (most?) grant-funded research groups exist only because they can function in that small team mode, with relatively thin budgets and the need to wear many hats.
The only safe way to scale work under these grants is what we see already: schools expand to have additional PIs running their own small, independent labs. These eventually compete with each other almost as much as they compete with outside labs. From the host school, they get actual "non-core" services like building facilities, human resources, payroll, contracts and legal support, etc. But, no one group can really scale to have more staff and specialization focused on a single research effort. There is a reason that "big science" projects are few and far between and are mired in much more politics---the granting agencies have to be motivated to write different kinds of funding opportunity to bootstrap such projects and to continue their operation. The budgets and timelines are vastly different.
It's why many PhDs can't get jobs in teams like mine - they don't have the skills of people with lesser degrees, and while we're not against training new candidates, we don't want to pay them large sums of money when the guy who just had a BS degree came in with all these skills.
They can also learn them quite fast before the interview/job, and it's a concern that they don't, when almost all other applicants do.
This isn't about skills development. It's about using people who already have these skills as engineering support labor for the research group while paying then 1/2 to 1/5th of what they'd make in industry doing equivalent work (that, again, they already know how to do!)
But I think there’s also a basic misunderstanding about what a T-shaped individual is and why they are sought after outside academia. It’s not because they can do a bit of everything, setting up the build system with one hand and designing the logo with the other. (That’s more what you’d call a jack of all trades.)
The reason T-shaped individuals are sought after in industry is that they are effective leaders. They understand enough about the problem domain(s) to identify meaningful problems and enough about the solution domain to help map out the solution space.
The reason you’d want more T-shaped people in machine learning research is that they will ask better research questions, not because they’ll design prettier logos.
> The reason you’d want more T-shaped people in machine learning research is that they will ask better research questions, not because they’ll design prettier logos.
Author here---fully agree with you. Man, that example with the logos is going to haunt me now...I just did not want to bog down the article with too many technical details. Of course there's immense value in moving between disciplines...that's a whole other discussion to be had (because IMO, academia does not necessarily have the right incentives for this either; research between disciplines X and Y often tends to be dismissed as being 'not enough X for Y' or 'not enough Y for X')
I read some anger and disenchantment in your comment. Feel free to reach out in order to get another perspective on your situation. In any case, without knowing the specifics of your situation, and being a more or less anonymous internet stranger providing unsolicited advice, I would suggest you don't do anything hasty. Getting a PhD can be useful in certain domains; at the same time, if you are not engaged and---dare I say it---happy with your overall situation, it might be time to move on.
Whatever you do: the world is big and, with a little bit of luck and foresight, there are additional opportunities to be found! Best of luck to you :-)
I do feel anger sometimes, but after all they're pleasant folks to work with, and I feel I shouldn't blame them just because they're born in richer places than me. I'd probably still finish my PhD :) But I do admire your resolution to become a professor, as I have been nothing but impatient with teaching other people things I understand. Your blog does have many insights that are rare in academia, with so many people rushing to get more papers published. I hope you'd have a successful career as a professor.
I come from a non-academic background myself (although being born in Europe certainly simplified things...I am mentoring students from India, for instance, and was shocked to learn about the high cost of even applying somewhere---but that's a different story), and have the feeling that academia is closed off against these sort of backgrounds. Despite academics trying very hard to think of themselves as being 'heterodox' and 'challenging the status quo,' most just follow along with the system. I don't want to assume a self-aggrandising position and believe that I alone can change the system, but maybe I can point out some issues at least...
(and thanks a lot for your kind words!)
There are good points already mentioned. I can add if you hire one person to write unit tests, another for graphics, another for models... then you suffer the overhead of coordinating them. Also, pure research work is often repetitive and drudgery, so being able to come up with cool logo or working on "uninteresting and unexceptional" tools that can be shared with other researchers can maintain the motivation.
Author here--thanks, that's a cool way of putting it! (plus, working on such tools, for instance, should be recognised by PIs in an ideal world)
Sure, for example, it's a waste to have a newly minted PhD set up CI systems. However, it's more of a waste to have them blocked waiting on someone else and then beholden to someone else every time there's an issue.
Researchers that can drop into the technology stack are great productivity multipliers for themselves and for their peers.
Academic types do their students a disservice when neglecting the development of such skills. Not everyone lands a professorship.
Also, didn't a bunch of these ML folks get undergrad CS degrees? Why should they be unable to do undergrad CS tasks?
This isn't about skills development. It's about using people who already have these skills as engineering support labor for the research group while paying then 1/2 to 1/5th of what they'd make in industry doing equivalent work (that, again, they already know how to do!)
I am saying that every student should be doing these tasks. The students without these skills should be learning them too.
Then no one is being taken advantage of. Then all of the students are more marketable. These skills are useful to academics too: successful junior faculty will one day need to stand up their own labs with their technical skills.
For one, the main benefit of a funded Ph.D. is that you don’t pay tuition. That’s a huge benefit that can total $40k or more per year before candidacy.
Secondly, you (should) get to travel and present your research around the world for free. During my Ph.D. I got to present our work in Portugal, Chicago, and Japan, all expenses paid. That’s not nothing for a young aspiring researcher, and not something my friends in industry got to do at 24.
Also, one thing that a lot of people don't know is that grad students in the USA don't have to pay FICA taxes on their stipends or their tuition remission. That also has to be figured in when comparing compensation.
No one does a phd for the coursework :)
> travel
This is worth perhaps $2K-$4K. Your industry friends could easily travel the world, and they wouldn't be stuck in a conference hotel for half their trip.
> FICA taxes
No! This isn't a good thing!!!
The employer pays half of these, and paying them entitles you to the US social safety net! Universities don't even give phd students the minimal benefits that every other employer in the entire country is required to subsidize 50%.
Stated another way: Denying any other employee a 50% match on state-provided retirement/disability insurance/unemployment insurance/medicare/etc. would literally be criminal.
Tuition also entitles you to access library resources, university software, campus gyms, sports facilities, child care, and university provided health services. I think your emoticon is implying your statement is tongue in cheek, but I mean.... they absolutely do though. If you don't consider taking courses a benefit to doing a Ph.D. then yeah, the value proposition of a Ph.D. program is quite lopsided. Tax-free tuition is an amazing benefit from my perspective.
> Your industry friends could easily travel the world, and they wouldn't be stuck in a conference hotel for half their trip.
You say stuck at a conference hotel, but I love conferences. They are so much fun and I always learn so much and meet amazing people. Moreover, what I said was "travel and present research". Not many employees who have been working for 1-2 years out of college get to present their work in front of an audience of peers and experts. It's great exposure and an amazing way to make connections and start building a career. My friends in industry were never given opportunities like that.
> paying them entitles you to the US social safety net!
I mean, you start paying into them soon enough -- I'm not really worried about my social security payout when I retire because I didn't pay into FICA when I was 24. The benefit of not paying FICA is that $30k goes further now when you need it. With Obamacare, healthcare isn't much of an issue anymore for grad students, and we got cheap/free healthcare at the university health services.
Look, all I'm saying is that when you say "$30k with marginal benefits" it makes it sound like a grad student is being compensated as a fast food employee. It's more like: $30k, and there are all kinds of benefits which may or may not appeal to you as an individual. If you want to make a ton of money go into industry. If you want a different experience do a Ph.D. You will (should!) come out of it with an education and a lot of open doors, some of which are permanently locked or vastly harder to open without the Ph.D. degree.
I know Ph.D. students who made almost six figures by doing highly paid research internships. Some advisors only care about demos, while others care about startups, and yet others care about how the work is presented to funding agencies. I know Ph.D. students who did most of their Ph.D. in a completely different city than their degree-granting institution. I know Ph.D. students who were highly pressured to publish, and Ph.D. students who had no pressure to publish. I know Ph.D. students who basically had no advisor, and Ph.D. students with 3 world-expert advisors. Some Ph.D. students are funded by the university, others are funded by large collaborative grants, or by their own fellowships, and yet others are funded by small industry grants their advisor got.
It's like if threads about being a software developer had comments like "at tech companies, you only get rewarded for closing tickets."
It would definitely be cheaper for my research group to outsource this work onto non-research staff.
Author here! I realise that my point is a little bit misunderstood. I am not saying 'Ah, let the PhD students do it.' I am saying that these sort of aspects should be recognised and rewarded, potentially influencing hiring decisions (for instance, hiring an SWE role instead of a Ph.D. student to fully support the lab).
Not necessarily an opportunity cost. Often they are paid in a more valuable commodity, i.e. visas.
But I'm writing from the perspective a grant reviewer. Systematically abusing immigrant labor.... well, it isn't exactly a more charitable interpretation. So, from my perspective, a bit of a moot point :)
(There's also OPT, btw, which is critical in the process, and H1B lottery chance, assuming a US Masters degree, was more than 50% last time I looked at it a couple years ago. O1 (and even EB1A/B) are not that rare for well-published PhDs, and EB2-NIW is quite easy for any PhD worth their salt, I'd say.)
You can master-out in the middle of a PhD program, which is frowned upon, but it is difficult to get funding if you directly apply for a masters degree.
Perhaps the solution is to decrease the negative cultural backlash against predisposed intention of leaving with a masters despite applying for a PhD, which is currently seen as some sort of betrayal of the adviser if you are going to have your grad students do boring work. It'll be a win-win deal in that case.
When I was in school, one applied to "MS/PhD" programs most typically. And all imaginable variants existed in terms of funding and likelihood of leaving with only a MS. Some schools would fund no one till they finish the MS stage, and others would fund everyone from the beginning (just give every accepted student a full fellowship for the first couple years). And also some schools were pretty notorious for kicking people out at the end of the MS stage, with a difficult exam at this step or not enough advisors to take them on for the PhD.
One positive side, though, is if you wash out of a PhD program at any later stage, you at least walked away with a Master's.
That being said this is from my experience and am unsure how it is at other schools.
After tenure, it seems like profs drop down to a manageable 4-7 students in the disciplines I studied in. You can do that on 1-2 grants a year.
As others have previously mentioned it's a bit weird, because actually teaching is neither taught nor really supported by most research universities. It's often required, but not all that advantageous to do well.
I guess it depends on how you view an academic institution, is it to produce good thinkers or produce good work, although similar seems like you approach it in different ways when approaching it from one way or the other.
I'm not even upset with professors, I think Assistant Profs do the most work I've ever seen anyone ever do, its absolutely insane the crazy workload they take on to get tenure.
The advantage now of T type approaches is that, as the source points out so clearly, it gives you flexibility to grow, shift fields, and collaborate efficiently.
The article mentions, for example, maintaining the group's build system and making project logos. Different in kind, no?
PhD students should not be recruited with the expectation that they will do devops, software engineering, and graphic design work. Or, if that's the labor they're doing, universities should maybe start paying them a fair wage for their labor.
Building something like genenetwork.org in 1994 was exciting R&D for the time. Setting up a CI/CD pipeline or making a project logo in 2022 isn't.
Fully agree with you there---and for me, that was never the point of the article.
> Building something like genenetwork.org in 1994 was exciting R&D for the time. Setting up a CI/CD pipeline or making a project logo in 2022 isn't.
Yes, at the same time, maybe setting up CI/CD is needed nonetheless in order to better manage a shared codebase. My point is that PIs should be aware of these things and also recognise this sort of 'foundational' work whenever it happens. Of course, if a lab needs complex DevOps, then of course they should hire someone _dedicated_ for the role.
My observations pertain to the kind of work that happens behind the scenes and is often unrewarded and unrecognised. As I said: academia is doing a disservice to the people that are willing to (to re-use the example you supplied) set up a CI pipeline for their project or teach others how to write code cleanly, etc. Of course, the best approach would be to have dedicated roles for dedicated tasks in research (and beyond).
This is true for professors at research universities in grant fields. Their job is to bring in grants. Teaching is of minor importance in those fields, to the point that they don't need you to teach at all if you're that bad at it. Other fields don't have much in the way of grants, so teaching is given considerably more weight at all but the wealthiest institutions.
Whether the economisation of higher education is a good thing is a different question, of course. But in the UK there seems to be a strong drive towards separating teching and research. And to be honest, it becomes increasingly difficult to be good at teaching and research at the same time. If done properly, either is a full time job.
Teaching is generally under-valued across the board - graduate students are expected to teach as part of their PhD work very often, sometimes paid extra for this. I've seen some unscrupulous departments making no sincere attempts to help students finish their PhD on time, then turn around and offer up scraps like hourly teaching "contracts" to teach their classes, after their PhD funding has run out.
Separating teaching and research, if it leads to both being equal peers, can make sense. I fear, however, that teaching will continue to be the un-favoured step-sibling in institutions which have research income. Research income and grants let you build a fiefdom of underlings, but teaching generally doesn't - one professor can teach a class of 200, and run a couple of tutorial sessions with a team of 8 post-grad students.
The scaling of teaching is very attractive to the university, but it doesn't get the professor people under them, to make them more important.
I've seen people be hired into "active teaching" academic roles who are manifestly incapable of teaching. Their "research track record" seemingly made up for their car-crash presentation to the department during the interview process. After the pile of (entirely foreseeable, by anyone at the presentation) student complaints flooded in, they ended up not needing to teach. There's definitely an implicit assumption that "teaching is easy, and anyone can do it" lingering around in a lot of departments.
I don't think we should entirely split research and teaching - it's very much possible to be T-shaped and good at both. Indeed, being able to engage a room full of tired students first thing on a Monday morning is a skill many academic presenters would benefit from having, purely in learning how to better communicate their research. Unless departments take a much wider, more holistic view of what is expected though, this won't be valued or change, as far as I can see at least.
The whole thing is a fad.
Unrelated but there are a lot of things like this in the startup world as well. For example, Elon keeps saying all you need to do is find out what arrangement atoms need to be in and then figure out how to do it. Yeah, no shit. Drug discovery is just sticking the right atoms together? Or software engineering is just pressing the right keys on the keyboard and Michael Phelps is just modulating the right muscles at the right time? It sounds more profound than it really is. It is also completely unactionable and impractical.
Historically, it's a technician's job.
Of course it's super useful to have someone on an experimental physics team who can personally machine metal, blow glass, and repair broken electronics, but generally you try to leave that to specialists.
The people who built CERN are not the same as the people who designed CERN are not the same as the people who did the research that made something like CERN plausible.
Your attitude seems to be "Well - ML, web design, graphics, original research, it's all computers so why not?"
There's a reason Peter Higgs didn't operate a concrete mixer, and that's because it wasn't his job.
It's the same in CS. This kind of work should be handed over to someone who can work on it full time, so researchers can get on with research full time.
That's not what I meant at all, though. I realise that my perspective might be specific to machine learning research: here, most papers involve implementations, prototypes, etc. My point could be boiled down to the following: if you have a candidate with fewer papers but _also_ some knowledge on how to, for instance, write clean code (or at least clean_er_ code), that candidate can be an asset _just as much_ as the candidate going into a Ph.D. with 10 first-author papers already. Both candidates cover different aspects of the research area.
I am saying that it's a good idea to look beyond `$num_papers` as a metric when hiring budding Ph.D. students, being well aware of their respective responsibilities. Current hiring practice in academia, as far as I can tell, completely ignores the skills _enabling_ excellent research. Earlier in the thread, someone mentioned this nice saying, which sums it up pretty neatly:
> The astronomers thought I wasn't that great an astronomer but a great programmer, and the programmers thought I wasn't a great programmer but a great astronomer.
One last thing I want to point out:
> It's the same in CS. This kind of work should be handed over to someone who can work on it full time, so researchers can get on with research full time.
Yes, I will definitely hire full time positions for this. The biggest groups are doing this already, probably because they realise that having a software engineer on board can help. I will apply the same considerations when hiring for such a role, trading 'depth' in favour of 'breadth' when it comes to their skill set, though. That strikes me as the right way of approaching this.
Generalizations can make for identification of useful patterns. Sometimes, the identification is pareidolia. Sometimes, it works. Tools of science and philosophy help us move that along.
That's not how they are defined in the article or the detailed definition it links to. The definitions there define traits of someone who may be able to increase the agility of a team depending on the team and what it is doing but the same could be said for most types of people yet not all are T-shaped and vice versa.
…according to some model of the work which may be entirely inaccurate, I think is the point the parent is making.
> There is a veritable zoo of characters of the alphabet, trying to capture the elusive properties of people out there in the real world. It should be understood that I do not even for a second think that one pithy letter is sufficient to describe a real person; all of these characterisations are merely hinting towards certain properties. As always in real life, there are many different shades on a spectrum to consider here
T-shaped is somewhat generalized because over time people like this become more. Overall they are great for the start of projects and long term because they do what is necessary to ship products and projects.
> Over time, one might think that a T-shaped person thus turns into a comb-like pattern. As far as I understand, this has not been studied yet by ethnographers. Maybe the term ‘serial experts’ would also been appropriate, except for its unfortunate similarity to the criminal world.
The author states T-shaped people become the glue or "scaffolding" for a successful project.
> Ts as multipliers. Throughout my career, the most impactful projects always had a T-shaped person onboard. This person would usually not be an expert in the subject matter, but would be able to provide the direly-needed scaffolding and foundation of a project that is all too often ignored in the initial phase, until it comes back later on with full swing to wreak havoc
In my experience, T-shaped people can and do become experts of product, these are people more willing to make things work and ship projects over just focusing on their own goals or field of study solely. T-shaped people become product experts as they have seen more areas or earlier parts of the whole.
T-shaped people can take a project from start to finish due to them being product/market experts from their involvement in more than just one area. Startups seem to operate and start best with T-shaped. In game development, good game projects and interactive projects especially Ts are prevalent.
In game development for instance, Valve prefers T-shaped employees, generalists in many areas but deeper knowledge in one more more areas, people willing to help ship.[1] Even if later you have to get more depth in certain areas like rendering, networking, animation, audio or even branding/marketing, Ts are the trails that form that can later be upgraded or refined if needed. Ts are the root of self-organizing teams and companies like Valve.
Valve has a good take on the balance for T-shaped people that help ship games [1]. These types area willing to wear different hats even if it isn't in the area they want to be their main focus. The mere understanding of those areas from experience in them, can make their main focus more finely tuned to work and mesh with the product.
> We value “T-shaped” people.
> That is, people who are both generalists (highly skilled at a broad set of valuable things—the top of the T) and also experts (among the best in their field within a narrow discipline—the vertical leg of the T). This recipe is important for success at Valve. We often have to pass on people who are very strong generalists without expertise, or vice versa. An expert who is too narrow has difficulty collaborating. A generalist who doesn’t go deep enough in a single area ends up on the margins, not really contributing as an individual
Too much specialization can be bad for projects and people, especially early on. Though by definition Ts would have a one or more deeper knowledge skillsets and can bounce between I-shaped and T-shaped without issue.
For startups, games, apps, content, even entrepreneurship or sole-proprietors to small companies, T-shaped is the way it has to be in most cases. Everyone is somewhat T-shaped, it is a sliding scale.
For academics at the Masters/Phd level, being T-shaped may actually be harmful or even hidden on applications because the nature of academics is the deeper I-shaped focus, you are trying to further the deeper detail of some subject. I bet lots of more broad generalists leave this off their descriptions/applications because of the bias against it. Everyone is T-shaped, just some are more willing to do different roles and wear different hats. It is a balance of being a leader in their chosen subject, but also wanting to get their foot in the door. A T-shaped person can be an expert still, and in many areas, but stating too much generalization may lead to bias in academia as this article summarizes.
[1] Valve Employee Handbook (page 46) https://steamcdn-a.akamaihd.net/apps/valve/Valve_NewEmployee...
The astronomer's thought I wasn't that great an astronomer but a great programmer, and the programmers thought I wasn't a great programmer but a great astronomer.
So maybe Stoll appeared as a T-shaped person to others but a line to himself?
[1] https://academia.stackexchange.com/questions/122426/how-many...
It's great for the PI if they have T-shaped students on their team, but they must make it clear to those highly talented people that academia is most likely not for them. Which puts in question whether it makes sense for them to do a PhD at all.
The problem is, the PI has hardly an incentive to tell these valuable, lowly-paid and highly motivated people that they made a bad career choice by starting a PhD.
That's completely true in my experience, and not only because of the "lack of depth". Interdisciplinary work is simply not rewarded, because your papers are only partially recognized in the department you apply to. I worked in a cog.sci/comp.sci mixture with the focus on language (which adds linguistics), but the psychology departments didn't accept CS conference publications. That my principal paper got a pretty decent citation score for that field, and that some of the follow-up papers including experiments didn't do bad either, didn't mean anything to the CS department: they simply don't care about the topic, and nor "Cognition," nor "Cerebral Cortex" featured on their impact lists.
That said, I think the article's examples (setting up a build system, designing nice mock-ups or logos, creating animations or graphics for presentations) are not T-shaped academic skills, at all. Creating presentations is something everyone has to learn, the rest is offloading support work on PhDs. I've got an idea: perhaps they can do the teaching as well. That's a good T-skill. And do lab support. And get coffee, because you know, so much breadth. No, that's pure exploitation of, exactly as you call it:
> valuable, lowly-paid and highly motivated people that they made a bad career choice by starting a PhD.
PhD programs should only be allowed when there's a guarantee of a worthy curriculum, and sufficient supervision, and support. There'll be less PhDs, but that might even be beneficial.
> PhD programs should only be allowed when there's a guarantee of a worthy curriculum, and sufficient supervision, and support. There'll be less PhDs, but that might even be beneficial.
Absolutely! Maybe that's the way forward.
(I replied to another comment of yours, but wanted to get this in as well) Yes, absolutely, and this is somewhat painful to me, because I think that a PhD can also be a period of tremendous joy for such people because they might use it to experiment broadly with their skill set. Academia loses out here :-(
Definitely! But in the end they come out frustrated, because they get overtaken by their more focused peers on the career ladder.
There are other opportunities aside from a PhD where talented people can hone their skills while earning comparably low salaries. Startups, for example.
I’ve heard a lot of companies explicitly recruit T-shaped individuals. But I’ve never heard the opposite.
>Setting up a suitable build system. Writing some unit tests for the code. Creating a skeleton of the models that are to be developed. Designing nice mock-ups or logos. Creating animations or graphics for presentations.
This makes me think the author is a fairly I-shaped person, because those examples look like one area (i.e., build systems, unit tests, sketching models).
An example of a T-shaped person would be someone who can develop wet-lab experimental technique as well as new machine learning models (which in academia you have the space to do).
(but you are spot on: the breadth of my skills depends on your Zoom level...I am only a lowercase t at best :D)
I guess for me it is assumed that one who does the work can also communicate the work in various forms (text, formal presentations, informal discussions). Perhaps it requires separate skills? But to me they sound like a continuation along the same "I".
When one hears a nice talk from (what I would call) an "I" researcher, it is usually communicated well, but the techniques and scientific question can be narrow.
When I hear a nice talk from a "T" researcher, it's usually someone who combines seemingly disparate fields (e.g. physics + sociology) together to solve a scientific question.
I guess either way I assume "I" and "T" researchers are both able to communicate their work. Perhaps that's where I misunderstood your post.
Doing wet-lab experiments and developing in-silico ML models are completely different and often times labs are often focused on only one or the other. I imagine a T-shaped person excels in wet-lab experiments to generate quality data as well as in analysis/modeling skills to put the data into a quantitative framework.
I also want to make it clear that I don't consider "I" or "T" any better than the others. We need both in science to progress. Perhaps our definitions of "I" and "T" differ slightly.
I was thinking more in terms of skills that translate between different projects, such as the large degree of software engineering skills that help in every ML project. I realise that this is a relatively narrow definition of 'T,' and yours is much, much larger.
Science is definitely benefiting from all types of researchers; I was just observing that some hiring committees ignore non-core aspects of research in CVs, focusing only on the publication output.
> I imagine a T-shaped person excels in wet-lab experiments to generate quality data as well as in analysis/modeling skills to put the data into a quantitative framework.
Couldn't agree more! And often, at least in my observation, it's better to have someone who is a 'jack of some trades' here...
> I also want to make it clear that I don't consider "I" or "T" any better than the others. We need both in science to progress. Perhaps our definitions of "I" and "T" differ slightly.
Absolutely :-) And thanks for pointing out these differences in definition! Goes to show that I am living in my own little bubble...
I was thinking about famous "I" and "T" scientists in history and I think there is a connection between "T" and "I" scientists and Freeman Dyson's "Birds and Frogs" essay.
Examples of Frogs or "I": Francis Bacon, Freeman Dyson, John von Neumann
Examples of Birds of "T": Descartes, Yuri Manin, Hilbert
Now for people like me:
Give me at least a million USD to invest in next PhD in CS. The most BS form of PhD is PhD in CS. It chirps your life away, with no clear benefits. It doesn't scale well enough in a field where innovation is rapidly progressing. Also when you are being underpaid while working for someone else's vision. You can't do what you want, because "supervisor" may not like it.
And once you come out of your PhD, who is going to hire you? Industry or academia? Well even in academia, there are tons of PhD already, so goodluck competing that.
And in industry goodluck finding that sweet industry that requires research.
I was always sold by the dream that PhD gives time and opportunity to involve in high leverage problems. Turns out it is a just another churn. What value does this paper bring? Nothing.
When most of valuable work is being done in industry and groups collaborating across nations. And to be in such group, you have to have background. (People whom I mentioned earlier)
Instead of wasting 5 yrs on CS Phd, invest on one single company. That will bring value x10 than stupid PhD.
Since you seem to be talking about a CS Ph.D. in particular, there are actually many open faculty positions because industry is able to absorb so many graduates. It's quite common for a CS Ph.D. to get an academic tenure-track position right after graduation, without having to do a post-doc. We are interviewing many such candidates right now at my department, and we have been running searches since before my time. I was hired this way actually. It's competitive, sure, but you won't be sitting around for 5 years like my friend in Biology.
Actually the most searches fail because it's really a buyer's market for applicants. They can negotiate huge startup packages by leveraging competing offers that our department can't even hope to match, and so the search continues...
However, most people don't view PhDs as a money-making endeavor. IMO, that's a good thing. And no, most of the important work does not happen exclusively in the industry. Academia is very valuable, just not for the same kind of stuff.
Let academia die the miserable death it deserves. Let us instead focus our efforts on building its replacements rather than participate in this research funding gangbang of society. Because, make no mistake, we are all paying for this garbage.
For replacements:
1. We need systems meant to EDUCATE goddamnit. (e.g. Khan Academy)
2. We need systems focused entirely on producing PUBLIC and cutting edge research where resources are assigned on merit. (Don't yet know what these research systems look like.)
No Child Left Behind was supposed to implement this; the problem that quickly became obvious is that schools that were already doing well got more money because their students had higher scores and schools that were struggling actually lost funding because their scores were garbage. This contributed to the gap widening instead of closing.
It's difficult to figure out how to develop a system where resources are allocated based on objective potential rather than historical metrics.
I do believe that academics have come up with the right approach to determine the value of research - anonymized peer review. The problem is, there are always signals that leak through that veil of anonymity that help you determine who the authors and even the reviewers actually are.
> Perhaps the most adverse unintended consequence of NCLB is that it creates incentives for schools to rid themselves of students who are not doing well, producing higher scores at the expense of vulnerable students' education. Studies have found that sanctioning schools based on average student scores leads schools to retain students in grade so that grade-level scores will look better (although these students ultimately do less well and drop out at higher rates), exclude low-scoring students from admissions and encourage such students to transfer or drop out.
https://www.npr.org/sections/ed/2015/10/27/443110755/no-chil...
> Ahn found a very different story among schools where lots of students were struggling. For these, often poorer schools, the law was like quicksand. Donna Brown is director of federal program monitoring and support for North Carolina's public schools, and she saw the quicksand firsthand.
> "When I came to the department in 2004," Brown remembers, "there were nine schools in the state that were identified for some level of improvement sanction. And, by 2008-9, there were 521."
The replacements you mention fall exactly into that category: depending on where you are working, your teaching/outreach/educational aspects might actually count against you when it comes to advancing your career in academia. That is certainly a problem.
The second point I don't entirely get---the grants I applied for all demand public research, for instance open-access publications, open source software etc. (European perspective, YMMV)
1. The grant system overly rewards proposals that grant givers can understand to have a high chance of success. This creates an echo chamber of what kind of research gets funded.
2. The acceptance of mafias based on mutual back-scratching. The dynamics of funding into these mafias.
3. The cut of research funding claimed by universities as operational costs. (How much of your grant money goes to your university as overhead?)
The French Academy of Sciences used to run a mathematical grand prix (in the 18th and 19th centuries... maybe not anymore?) where they would put out a call for essays about a particular problem (e.g. the three body problem) and award a substantial prize to the winning essay. I like that kind of setup more, and I think there could be room for a parallel funding system where private companies or groups of interested individuals offer funding for targeted research using this kind of grand prix methodology.
I was not criticizing academic research for not happening in the public, was merely stating that that should also be a property of the new system.
Also, the same institutions should not be responsible both for education and for research. The objectives are clearly not aligned.
> 1. The grant system overly rewards proposals that grant givers can understand to have a high chance of success. This creates an echo chamber of what kind of research gets funded.
This is indeed something I am experiencing across the board. There are few 'blue-skies research programmes.' The Swiss National Science Foundation had one a few years back and I was over the moon to get a grant there...because I could not really guarantee success.
I am still somewhat struggling when writing a proposal: on the one hand, I want to show that I considered all the angles that can be considered; on the other hand, it's research, so I probably did not foresee all the problems that might arise. Getting this balance 'just right' is a knack, it seems.
> 2. The acceptance of mafias based on mutual back-scratching. The dynamics of funding into these mafias.
I never experienced this so far, but I can definitely see that 'networks help.' When you are in the in-group, everything is a little bit easier.
> 3. The cut of research funding claimed by universities as operational costs. (How much of your grant money goes to your university as overhead?)
OK, that one hit close too home! I just had exactly this discussion with someone from our own admin team. They are doing an excellent job, but they are getting quite miffed by the overheads claimed from the grant admin people (a different division). The numbers are even bigger outside Europe, or so I am told...
I am definitely with you in that the system needs improving. At this point, I am still somewhat optimistic that I can at least try to enact some changes by playing the game according to my own rules, but I am not sure whether reality is going to catch up with me at some point, overriding my 'idealistic naïveté generation field.'
I have also spent almost a decade in and around the US healthcare system and similarly think that it is impossible to improve.
The thing is, there are enough people at the fringes of both those systems who genuinely believe in the ideals of those systems. I think those people, working together, can build better institutions that exist in parallel to the corrupt, diseased, broken systems that are a burden on our society today.
For example, the T-shaped people who are undervalued by academia but still choose to focus their efforts there rather than in industry.
I am seeing flickers of this at least in some parts of academia, for instance the move towards open-access publishing and improved reviewing...but in the end, I have to admit that there might be a 'brain drain' coming (or people might, over time, just be corrupted and play along with the status quo)