Rambles around computer science
Diverting trains of thought, wasting precious time
Wed, 26 Feb 2020
Postdoc follow-ups
[I'm on strike again at the moment, just as when I wrote my last higher-ed piece, to which this is a follow-up.]
My last higher-ed piece, about postdoc myths was read rather more widely than I expected. (Thanks for reading!) That has left me with a few things to clear up, and a few follow-up thoughts which I didn't get on to last time.
Firstly, let me qualify: my take on postdoccing is more than a little UK-centric, and certainly doesn't generalise in all possible directions. However, I do believe it generalises to many places outside the UK, perhaps in non-obvious ways. The most contentious question of generality (at least in the Hacker News discussion) was whether postdocs “formally exist”. I gathered that many US institutions offer roles like “Postdoctoral Scholar”, for example. But my point was more about how the regulations of institutions and of funders haven't adapted. Job titles are at best a weak indicator of this, and to see jobs advertised as “postdoctoral X” is not enough to infer that there is any recognised status of “postdoc” in the institution or the wider academy, beyond “paid lackey”. Even in the UK, we see jobs advertised, including at the University of Cambridge, with titles like “Postdoctoral Research Associate”. That doesn't mean the institution has any official position of “postdoctoral” anything; it doesn't. The word is simply added for illustration; it is formally meaningless. Such employees' academic standing has been more accurately summarised as “people who do not exist” (to borrow a phrase from Anthony Edwards's remarks on the history of such positions at Cambridge). The high-level point is that institutions' and funders' processes are not designed around present career structures—where one might spend an unbounded number of years as a qualified researcher of considerable independent potential but not holding a “full” “academic” “position”, however that might be recognisable locally. Advertised job titles are not a good guide to reality.
For the same reason, it's wrong to suppose what's happening is “higher supply leading to lower price”. I've been talking about a degradation of the offering—early-career research jobs being offered on shorter contracts with fewer rights and less institutional status—and it's appealing to suppose this degradation is the result of “universities extracting more value” from the labour pool. But that is factually wrong. Neither pay nor status is re-negotiated on the basis of changing supply. Pay scales are hard to change; university regulations are even harder. To redefine positions at lower pay or lower status is a political act; someone has to pull the trigger on it. That isn't what has happened. Equally, in those cases where we would expect upward pressure we also don't see upward changes: universities and academics often find it difficult to hire postdocs with certain skills they want, but that rarely creates any action to improve pay and status (beyond a regulation-limited amount of salary-bumping), because the relevant political change is mostly beyond the means of the academics who are hiring. A key example is that many institutions face a chronic difficulty in hiring research software engineers. As far as I know, this hasn't driven many universities to reform their regulations. Instead, they have shown a boundless capacity simply to limp along with the problem uncorrected. For the same reason, there's no reason to believe downward pressure actually has much effect in cases of oversupply.
So if it is not a case of rational decision-making by universities in the face of increased supply, what is causing the body of underpaid under-statused researchers to get larger? In the UK at least, the answer is simple: it's the government, stupid. What we've seen is that the relative occupancy of pre-existing pay and status levels has been changing. That change arises not from the dynamic between universities and the labour market, but from that between universities and government. It's not supply and demand; it's poorly chosen public policy, formulated by ministers and civil servants who (as far as I can tell) don't understand research. What does change far more easily than pay-scales and regulations is budgets—what government controls. Hence the degradation is arising indirectly, not via the labour-market mechanism but by external changes to distribution of money between streams, and hence of people among the distinct scales and regulations that those pots feed. In short: for a given level of spending, we are relatively funding more postdocs and relatively fewer “full” academic staff. Note, as I argued last time, it would be wrong to equate the latter with “permanent” positions (or even with “teaching” positions). Note also, as I'll return to, the problem is emphatically not one of “not enough money”.
Career-wise, what were once stopgap arrangements—“spend a couple of years on this contract before a ‘proper’ academic role comes around”—have, creepingly, become the norm. Longstanding regulations and arrangements for “contract research staff” are applied increasingly far beyond their originally conceived uses, to an ever-larger and more ill-fitting body of staff, and over longer durations for each individual. But from the universities' point of view this is a case of boiling frogs, not rational agents. Meanwhile, I don't believe government is doing this deliberately; they're just asleep at the wheel. In fact, they don't realise they have the wheel. The (limited) evidence I've seen, such as government's response to the damning 2002 report of the House of Commons Science & Technology Committee and more recently then-minister Chris Skidmore's confused remarks on the subject (Hansard; tweet with video), is that government imagines it has no role in this, and it's all the universities' doing. But universities' hands are largely tied by how money is delivered. Of course, top-end institutions including Cambridge are culpable for their complacency in failing to challenge government.
Those two streams are “core funding” and “project funding”, which in the UK are known as the “dual support” system. I have a draft of a working paper on this subject, which I wrote as an assignment for a module in my PGCHE. I am hoping to expand it into something publishable; comments are very welcome, but be aware it is very much a draft at present. It is, necessarily, very UK-specific. It argues that the changes have come about indirectly, as unintended consequences of well-intentioned (but misguided) policies going back at least as far as 1981 and the attempt to “protect science” from wider public spending cuts. Later changes, to do with funding capital costs (“sustainability”) and with fairness and transparency of funding allocation (“selectivity”) have exacerbated the problem. The foul icing on the horrid cake is a lurking confounding variable—how much core funding is de facto being used to match-fund project grants that are under-costed.
This latter effect is subtle, and is the aspect most in need of further research. Although the headline data is clear that the block/project split has flipped from 60:40 to 40:60 between 1978 and 2018, the reality is almost certainly more drastic than that because more of the block grant is used as “match” or “top-up” support for the increasings volume of projects that are funded at below full economic cost. My lone data point so far (detailed in the draft article) is that in Cambridge, nearly all of the block research funding is being spent on subsidising project funding, i.e. on allowing it to continue being costed below the full economic rate. That's something my future research must dig into, along with a cohort-tracking study of the pre-92 universities to separate out the effects of debinarification in the early 1990s. To make clear statements about career progression, it'll also be necessary to make corrections for rank inflation: early indications are that it's now easier to get to [full] Professor, but no easier to get to lecturer [a.k.a. Assistant Professor], with consequences for how spending is distributed. Figuring out how much this generalises beyond Cambridge is another goal; my article does include some study of Kent, but so far it's less conclusive. If anyone knows another UK pre-92 university that publishes (or makes available to researchers) good-quality data about its staffing and spending over the past decades, please let me know.
The final thing to remember is that real-terms government spending on research has gone up considerably. Therefore, it's doubly unforgivable that career structures are in such a mess. When people like Sir Leszek Borysiewicz say “we don't have money to create better positions”, they're either ignorant or lying. The scarcity is entirely artificial, created by how the increased spending has gone disproportionately on project funding. This is both directly harmful (projects in themselves are a poor basis for both research outcomes and for careers), and indirectly harmful (projects, being under-costed, soak up additional block funding).
To sound a note of optimism, there are multiple ongoing shake-ups of UK government research funding. One is the mooted creation of an ARPA-like agency. Another is the “rebalancing to the regions” which suggests a brake on various institutionwise preferential attachment effects (discussed in my previous post) that have harmed career structures under the project-dominated funding regime. Both of these shake-ups are being driven by Dominic Cummings—a dislikeable figure to put it mildly, but one whose influence may yet do good in this space. At the recent Research Institutes' Conference organised by the National Cybersecurity Centre, the panel session involved three videos, one of which featured Cummings quoting Alan Kay's dictum that we should “fund people not projects”. I think Kay is exactly right, but it's interesting how often his words are misunderstood, and unclear whether Cummings has understood them. In a later post I'll continue this discussion with some notes on how this can go wrong.
[/highered] permanent link contact
