The suggested hiring process, especially solving work related tasks while thinking out loud, sounds a lot like how software developers are interviewed.
I wonder to what extent we could untangle bench work from cognitive work, and how little context overlap is needed between who designs the experiment and who carries it out. Similarly to how coding agents (claude code, cursor...) have increased software developers' productivity, allowing postdocs to spend significantly more time on the cognitive parts of their job could lead to them asking better questions, devising more efficient ways of testing them, and being better at analysing and sharing their results.
In biotech and pharma this structure seems to be working – I wonder if academia could reimagine its funding and career systems to follow industry's example?
I think deep knowledge of physical techniques is probably required to design experiments well. With coding everything is fully deterministic and exact, but with wet lab science, two scientists could follow the same algorithm (a protocol) and get very different results depending on skill.
That said, you don’t have to be actually good at enacting protocols if a robot can do them for you. In this instance though, robot programming/prompting becomes a new scientific technique that requires mastery.
I agree re academia shifting career systems. It’s just so stagnant and dogmatic due to lack of incentive to change!
Completely agree with this! Been talking about something similar with friends for quite a while. Science - bench work - is so practically driven, but when we start at somewhere like a PhD we’re woefully under equipped. Lots of science writers I know bailed on research because the practical aspects simply weren’t for them (for lack of skill or interest reasons).
Would love to see something more akin to a Lab Apprenticeship, moving graduates from their uni lectures straight into lab work to get a handle on just how practical it is, before diving into a PhD. It is blue-collar work for a large part and learning experimental design from a hands-on perspective would sound wonders for prospective scientists’ confidence and general aptitude at the bench.
Exactly! Acknowledging this reality and shifting how we frame science would also likely attract slightly different people to the career path—probably for the better
The concept of lab techs exists and those don’t have to be the PhD scientists. If it’s PhDs doing all the pipetting then it may be because there is an oversupply of mediocre PhDs. Everyone wants maximum status. Everyone wants to be an intellectual, at least upper middle class, educated white collar. It’s the mantra everywhere. Nobody dreams of sending their kids to one day robotically move lab material and vials from one fridge to another on the command of a boss. That’s not prestigious enough. Make it dignified by the culture and they will be more willing. And or pay better. But again, if there is an oversupply, they don’t have to. And robotics will make all this go away pretty soon either way.
"there is an oversupply of mediocre PhDs" -> exactly correct. I have an essay in the works on how robotics will solve much of our angst surrounding academia and its power structures. Not sure many have priced this in or thought about it, but it will be glorious.
In my view the real function of academia is to attach prestige to ideas. If the current iteration is somehow obviated by AI and robots, there will still be demand for producing people who have the necessary status to rubberstamp ideas with the prestige of centuries of intellectual history backing them up.
And just like a blue collar vocation such as plumbing or being an electrician, recent grads should have apprenticeships, not internships! Long-term, paid, structured programs to learn the skills of the trade from an expert. Not a few months of unpaid or minimally stipended work that is less hands-on or strictly observational. I’m talking 1-5 years of truly learning and honing your craft🤘
Something similar to this, which I imagine could be useful in e.g. hospital labs (and any lab that uses a pretty limited set of protocols) would be to hand them one of the standard protocols, have them show up, and just let them at it. It's a bit less of a time investment, since you're not shilling out a whole day plus whatever time it takes for the previous step where they have to make their own protocols.
Not relevant for every lab, of course, but a somewhat simplified version of your idea, that still shows their actual practical skills.
Absolutely. You're hardly going to attract the best people with mediocre wages! I think part of the way to solve that is to reduce the total number of people getting PhDs by increasing the rigour. Currently the same pot of grant money is spread across far too many scientists and administrators for wages to be competitive.
I largely agree with this, though the most competent lab techs I've seen at the bench tend to still have college degrees and would likely score high in trait conscientiousness. They also usually were mentored heavily for some meaningful duration.
When we say "blue-collar" this usually means a job that doesn't require a college education. Whether our not lab work requires the filtering function of requiring a college degree likely depends on how much independence the individual will be given on the job, how much experimental design/project management work is required, how much reading/writing/analysis is required, and how sophisticated the related bench protocols are.
I've also felt that too much is bundled into various roles in academic science, and intellectual independence is often romanticized in unproductive and inefficient ways. It varies substantially by PI and lab resources. Most of my experience is from a large large (always 20+ with multiple grad students, post docs, techs, staff scientists, undergrads and med students rotating through), where almost every individual was tasked with pioneering some independent, primary research work excepting some technical stuff with otherwise very narrowly defined roles. Some of the problems of organization are a function of what the desired work product is: publications.
True. I agree that we shouldn't necessarily eschew the knowledge-base gained from an undergraduate education (although the degree to which this actually happens these days is questionable ...).
I'd say the college education is necessary but not sufficient to be a good scientist in my opinion. Excellent technical skills are another necessary but not sufficient component (and under-emphasised--hence my essay). Higher level experimental design, ideation, creativity etc are also required. All quite separate things and they complement each other synergistically (multiplicatively, not additively). In my view, all of these components should be contained within a single human ideally. Perhaps that's where we disagree? You're saying you prefer the compartmentalised lab where each member specialises in their specific function? I can see that might be more efficient in some ways. I'd say it swells payroll unnecessarily and loses out on the crosstalk that can happen within one person's mind if they are acutely aware of experimental details and methods while designing experiments or while coming up with project ideas etc. Interpersonal communication is going to be less efficient than just thoughts within the same mind.
It's definitely true, though, that intellectual independence should be a rare commodity given only to the most creative and competent scientists. And for everyone to have that as a goal is a recipe for disappointment perhaps.
I'd say the college education is necessary but not sufficient to be a good scientist in my opinion. Excellent technical skills are another necessary but not sufficient component (and under-emphasised--hence my essay). Higher level experimental design, ideation, creativity etc are also required. All quite separate things and they complement each other synergistically (multiplicatively, not additively). In my view, all of these components should be contained within a single human ideally. Perhaps that's where we disagree?
I don’t disagree per se. I just see some issues with how many labs are currently staffed and led. The organization of many important labs is more geared toward optimizing the number of grants/papers a PI can own rather than answering challenging and worthwhile scientific questions or developing new and effectives scientists. Specifically, there is an issue with many large, sometimes very prestigious labs rarely having formalized organizational structures and defined responsibilities that ensure the transfer of explicit and tacit wet lab knowledge/skills to new team members. Is it good that we’ve turned some many PIs into paper pushing admins who mostly review the work that goes on in their labs at the level of final work products (presentations, abstracts, grants, papers, etc)?
You're saying you prefer the compartmentalised lab where each member specialises in their specific function? I can see that might be more efficient in some ways. I'd say it swells payroll unnecessarily and loses out on the crosstalk that can happen within one person's mind if they are acutely aware of experimental details and methods while designing experiments or while coming up with project ideas etc. Interpersonal communication is going to be less efficient than just thoughts within the same mind.
It’s not compartmentalization so much as professionalization into clearly defined management structures that are responsive to the turnover inherent to training scientists. Some of this means having a technical staff that does very specific things only while the scientists in training network with and progress through the lab comprehensively.
It's definitely true, though, that intellectual independence should be a rare commodity given only to the most creative and competent scientists. And for everyone to have that as a goal is a recipe for disappointment perhaps.
Yes, so many grad students and post docs start with a fairly vague sense of what their “project” or “research question” will be and are charged with figuring everything out on their own. But the “smart” grad student/post doc in this situation will just optimize on whatever will optimize their chance of specific career advancement without regard to whatever is scientific valuable or would provide the most value within the lab’s strengths and limitations.
Yeah sorry I should’ve substituted “science” for “life sciences” throughout. I don’t know much about the day in the life of an astronomer, but I imagine the same principle applies. The point is to make interviews more reflective of the actual work of a scientist.
The suggested hiring process, especially solving work related tasks while thinking out loud, sounds a lot like how software developers are interviewed.
It just seems like a sensible way of doing things. Software engineers know a thing or two!
Very much agreed!
I wonder to what extent we could untangle bench work from cognitive work, and how little context overlap is needed between who designs the experiment and who carries it out. Similarly to how coding agents (claude code, cursor...) have increased software developers' productivity, allowing postdocs to spend significantly more time on the cognitive parts of their job could lead to them asking better questions, devising more efficient ways of testing them, and being better at analysing and sharing their results.
In biotech and pharma this structure seems to be working – I wonder if academia could reimagine its funding and career systems to follow industry's example?
I think deep knowledge of physical techniques is probably required to design experiments well. With coding everything is fully deterministic and exact, but with wet lab science, two scientists could follow the same algorithm (a protocol) and get very different results depending on skill.
That said, you don’t have to be actually good at enacting protocols if a robot can do them for you. In this instance though, robot programming/prompting becomes a new scientific technique that requires mastery.
I agree re academia shifting career systems. It’s just so stagnant and dogmatic due to lack of incentive to change!
Completely agree with this! Been talking about something similar with friends for quite a while. Science - bench work - is so practically driven, but when we start at somewhere like a PhD we’re woefully under equipped. Lots of science writers I know bailed on research because the practical aspects simply weren’t for them (for lack of skill or interest reasons).
Would love to see something more akin to a Lab Apprenticeship, moving graduates from their uni lectures straight into lab work to get a handle on just how practical it is, before diving into a PhD. It is blue-collar work for a large part and learning experimental design from a hands-on perspective would sound wonders for prospective scientists’ confidence and general aptitude at the bench.
Exactly! Acknowledging this reality and shifting how we frame science would also likely attract slightly different people to the career path—probably for the better
The concept of lab techs exists and those don’t have to be the PhD scientists. If it’s PhDs doing all the pipetting then it may be because there is an oversupply of mediocre PhDs. Everyone wants maximum status. Everyone wants to be an intellectual, at least upper middle class, educated white collar. It’s the mantra everywhere. Nobody dreams of sending their kids to one day robotically move lab material and vials from one fridge to another on the command of a boss. That’s not prestigious enough. Make it dignified by the culture and they will be more willing. And or pay better. But again, if there is an oversupply, they don’t have to. And robotics will make all this go away pretty soon either way.
"there is an oversupply of mediocre PhDs" -> exactly correct. I have an essay in the works on how robotics will solve much of our angst surrounding academia and its power structures. Not sure many have priced this in or thought about it, but it will be glorious.
In my view the real function of academia is to attach prestige to ideas. If the current iteration is somehow obviated by AI and robots, there will still be demand for producing people who have the necessary status to rubberstamp ideas with the prestige of centuries of intellectual history backing them up.
And just like a blue collar vocation such as plumbing or being an electrician, recent grads should have apprenticeships, not internships! Long-term, paid, structured programs to learn the skills of the trade from an expert. Not a few months of unpaid or minimally stipended work that is less hands-on or strictly observational. I’m talking 1-5 years of truly learning and honing your craft🤘
Superb idea.
Something similar to this, which I imagine could be useful in e.g. hospital labs (and any lab that uses a pretty limited set of protocols) would be to hand them one of the standard protocols, have them show up, and just let them at it. It's a bit less of a time investment, since you're not shilling out a whole day plus whatever time it takes for the previous step where they have to make their own protocols.
Not relevant for every lab, of course, but a somewhat simplified version of your idea, that still shows their actual practical skills.
Great suggestion!
For bench scientists this is absolutely true...as one moves gradually moves from the bench to the boardroom, this gradually becomes less true...
That's fair. Different jobs though! I'm not sure I'd call both of them "scientists" in the true sense
Perhaps they should also get paid more
Absolutely. You're hardly going to attract the best people with mediocre wages! I think part of the way to solve that is to reduce the total number of people getting PhDs by increasing the rigour. Currently the same pot of grant money is spread across far too many scientists and administrators for wages to be competitive.
I largely agree with this, though the most competent lab techs I've seen at the bench tend to still have college degrees and would likely score high in trait conscientiousness. They also usually were mentored heavily for some meaningful duration.
When we say "blue-collar" this usually means a job that doesn't require a college education. Whether our not lab work requires the filtering function of requiring a college degree likely depends on how much independence the individual will be given on the job, how much experimental design/project management work is required, how much reading/writing/analysis is required, and how sophisticated the related bench protocols are.
I've also felt that too much is bundled into various roles in academic science, and intellectual independence is often romanticized in unproductive and inefficient ways. It varies substantially by PI and lab resources. Most of my experience is from a large large (always 20+ with multiple grad students, post docs, techs, staff scientists, undergrads and med students rotating through), where almost every individual was tasked with pioneering some independent, primary research work excepting some technical stuff with otherwise very narrowly defined roles. Some of the problems of organization are a function of what the desired work product is: publications.
True. I agree that we shouldn't necessarily eschew the knowledge-base gained from an undergraduate education (although the degree to which this actually happens these days is questionable ...).
I'd say the college education is necessary but not sufficient to be a good scientist in my opinion. Excellent technical skills are another necessary but not sufficient component (and under-emphasised--hence my essay). Higher level experimental design, ideation, creativity etc are also required. All quite separate things and they complement each other synergistically (multiplicatively, not additively). In my view, all of these components should be contained within a single human ideally. Perhaps that's where we disagree? You're saying you prefer the compartmentalised lab where each member specialises in their specific function? I can see that might be more efficient in some ways. I'd say it swells payroll unnecessarily and loses out on the crosstalk that can happen within one person's mind if they are acutely aware of experimental details and methods while designing experiments or while coming up with project ideas etc. Interpersonal communication is going to be less efficient than just thoughts within the same mind.
It's definitely true, though, that intellectual independence should be a rare commodity given only to the most creative and competent scientists. And for everyone to have that as a goal is a recipe for disappointment perhaps.
I'd say the college education is necessary but not sufficient to be a good scientist in my opinion. Excellent technical skills are another necessary but not sufficient component (and under-emphasised--hence my essay). Higher level experimental design, ideation, creativity etc are also required. All quite separate things and they complement each other synergistically (multiplicatively, not additively). In my view, all of these components should be contained within a single human ideally. Perhaps that's where we disagree?
I don’t disagree per se. I just see some issues with how many labs are currently staffed and led. The organization of many important labs is more geared toward optimizing the number of grants/papers a PI can own rather than answering challenging and worthwhile scientific questions or developing new and effectives scientists. Specifically, there is an issue with many large, sometimes very prestigious labs rarely having formalized organizational structures and defined responsibilities that ensure the transfer of explicit and tacit wet lab knowledge/skills to new team members. Is it good that we’ve turned some many PIs into paper pushing admins who mostly review the work that goes on in their labs at the level of final work products (presentations, abstracts, grants, papers, etc)?
You're saying you prefer the compartmentalised lab where each member specialises in their specific function? I can see that might be more efficient in some ways. I'd say it swells payroll unnecessarily and loses out on the crosstalk that can happen within one person's mind if they are acutely aware of experimental details and methods while designing experiments or while coming up with project ideas etc. Interpersonal communication is going to be less efficient than just thoughts within the same mind.
It’s not compartmentalization so much as professionalization into clearly defined management structures that are responsive to the turnover inherent to training scientists. Some of this means having a technical staff that does very specific things only while the scientists in training network with and progress through the lab comprehensively.
It's definitely true, though, that intellectual independence should be a rare commodity given only to the most creative and competent scientists. And for everyone to have that as a goal is a recipe for disappointment perhaps.
Yes, so many grad students and post docs start with a fairly vague sense of what their “project” or “research question” will be and are charged with figuring everything out on their own. But the “smart” grad student/post doc in this situation will just optimize on whatever will optimize their chance of specific career advancement without regard to whatever is scientific valuable or would provide the most value within the lab’s strengths and limitations.
Eh… astronomy/physics is nothing like this. We don’t pipette. Are we not scientists?
Yeah sorry I should’ve substituted “science” for “life sciences” throughout. I don’t know much about the day in the life of an astronomer, but I imagine the same principle applies. The point is to make interviews more reflective of the actual work of a scientist.
On this I fully agree