Perverse Incentives? How Economics (Mis-)shaped Academic Science

By Ken Chiacchia, Senior Science Writer, Pittsburgh Supercomputing Center

July 12, 2017

The unintended consequences of how we fund academic research—in the U.S. and elsewhere—are strangling innovation, putting universities into debt and creating numerous PhD graduates and postdoctoral fellows who will not be able to get jobs in their chosen fields, according to economist Paula Stephan of Georgia State University.

The good news, Stephan said at the opening plenary session of the PEARC17 conference in New Orleans on July 11, is that researchers probably needn’t go back to the politicians to ask for more money. The bad news: the current system is so ingrained it’s hard to be optimistic.

“I don’t think it would take more funding to [encourage] more risk,” she said, but “unless we change the incentives in the system we’re going to continue to overbuild and over train.”

Stephan identified three major effects of the perverse incentives governing academic research: over-training, risk aversion and over-building of physical infrastructure. All three are problems in their own right but also feed back to make the situation worse.

“Economics is about incentives and cost,” Stephan explained, and both are problematic in most national funding systems. She particularly examined that of the U.S.

The Inaugural Practice and Experience in Advanced Research Computing (PEARC) conference—with the theme Sustainability, Success and Impact—stresses key objectives for those who manage, develop and use advanced research computing throughout the U.S. and the world. Organizations supporting this new HPC conference include the Advancing Research Computing on Campuses: Best Practices Workshop (ARCC), the Extreme Science and Engineering Development Environment (XSEDE), the Science Gateways Community Institute, the Campus Research Computing (CaRC) Consortium, the Advanced CyberInfrastructure Research and Education Facilitators (ACI-REF) consortium, the National Center for Supercomputing Applications’ Blue Waters project, ESnet, Open Science Grid, Compute Canada, the EGI Foundation, the Coalition for Academic Scientific Computation (CASC) and Internet2.

Over-training: A Plague of PhDs

Increasingly, Stephan argued, universities are following a “high-end shopping mall” model in which they “lease” space to researchers—the “stores.” Physical building, particularly during the funding increases of the 1990s, became a priority as universities vied to attract top-performing (read: highly funded) research faculty. One down side to this model, though, is that individual principal investigators took on so much of the risk. With about 95 percent of research faculty paying their own salaries through soft money, funding has become existential and devours increasing amounts of the average lab head’s time: One study estimated that PIs spend 42% of their professional time on grant administration and writing.

“This raises the issue of how you’re going to staff your lab,” Stephan said. While few researchers make a conscious decision to bias hiring toward some types of research workers, the economic pressures often give little choice.

The issue is stark in the decision of whether to employ graduate students, postdoctoral fellows or staff scientists to conduct lab research. Nationally, graduate students average a stipend of about $26,000 annually; in addition, they represent approximately an additional $16,000 or more for tuition and other student costs. Their hourly “pay rate,” then, can be between $19.50 and $27.50.

Postdoctoral fellows are paid more. But they also have no tuition costs and at most universities have few additional benefits. Assuming a university follows the NIH benchmark of $43,692 for a first-year postdoc, their hourly rate comes to around $17 to $18, depending on the field.

Staff scientists start at about $60,000 to $75,000, coming out to an hourly rate of about $30.00. But that doesn’t reflect their full cost, which includes much more extensive benefits than students or postdocs.

Given this incentive structure, Stephan explained, it isn’t hard to understand the relative scarcity of staff scientists. Her own study found that at least 72 percent of academic research papers had postdocs or grad students as their first author. In the NSF’s annual survey, life science PhD graduates with definite job commitments have fallen from a peak of 70% in 1994 to 58% in 2014—and most of those are going to postdoc positions, not permanent jobs.

With the scarcity of permanent positions for these postdocs to go to next, “academe has become the alternate career track” for PhDs, particularly in physics and the physical and life sciences, she said.

“Training [has become] less about the future supply and more about getting research and teaching done now,” Stephan said.

Aversion to Risk

Along with the oversupply of PhDs, the funding structure has created an atmosphere in which risk-taking is discouraged in the funding process. In an influential Proceedings of the National Academy of Science USA paper, biomedical giants Alberts, Kirscher, Tilghman and Varmus criticized biomedical research funding as overly risk averse. Researchers have perceived a similar problem in the physical sciences: Even DARPA, which once self-identified as funding risky projects, has been criticized for being over-cautious.

At the stage of grant application reviews, the common requirement for preliminary data among many reviewers tilts the field against high-risk projects. So does the use of bibliometric measures of author impact. The short-term nature of the funding cycle also discourages novelty: “It’s hard to recover from failure in three years,” Stephan said. And since the success rate for grant continuations is higher than that for new grants, the system encourages researchers to “stay in their lanes.”

“The stress on ‘translational’ outcomes” that provide immediate practical applications “also discourages risk,” she added.

Another study showed that highly novel papers tend to show pronounced payoffs at 13 years after publication but little at three years. Non-novel papers, on the other hand, pay off better at three-year cycles—but don’t improve over time.

If You Build It, They Will Not Necessarily Come

Overbuilding—the construction of unneeded university brick and mortar—came with the NIH budget doubling in the late 1990s. Universities, assuming continued growth, embarked on a “building binge” to attract top grant-attracting faculty. They borrowed to do so, partly because interest payments for debt service can be included in calculating indirect costs charged against those grants—and thus it would, presumably, “pay for itself.”

From 1988 to 2011, biomedical research floor space at the average university increased from 40,000 square feet to 90,000 square feet.

When funding declined in real dollars, unrecoverable debt and even facility mothballing followed. The annual average university debt service grew from $3.5 million in 2003 to $6.9 million in 2008. It created an economic drag on many research universities that will be hard to escape.

“All disciplines will pay for this, not just the biomedical sciences,” Stephan said.

The Way Out?

The irony, of course, is that the primary justification for government-funded research is to take risks that industry can’t.

For economists, the case for academe starts with a concept called “market failure,” Stephan explained. It’s the term used to describe the way most firms avoid overly risky projects; the difficulty of capturing financial benefits from fundamental discovery is a particular disincentive to pursue that which does not pay off in the near term.

“But the risky stuff shifts knowledge frontiers, eventually contributing to economic growth,” she said.

Excellence is not the same thing as risk-taking, Stephan took pains to add. Not all excellent research takes big risk; not all risky research is of high quality.

“I think as a country we need a portfolio,” she said. “It does not mean that there is not a substantial role for what we call ‘normal’ research.” But if we don’t change the incentive structures of our funding process—rewarding outcomes over longer time periods, creating incentives to encourage permanent rather than temporary jobs and make living on “soft money” less precarious, we won’t see the kind of innovation in which academic research was supposed to specialize.

“I’ve been working at this for too long, so I’m not wildly optimistic,” Stephan admitted.

Ken Chiacchia, Senior Science Writer, Pittsburgh Supercomputing Center, is following a non-traditional career path for science PhDs.

Subscribe to HPCwire's Weekly Update!

Be the most informed person in the room! Stay ahead of the tech trends with industy updates delivered to you every week!

SC Bids Farewell to Denver, Heads to Dallas for 30th

November 17, 2017

After a jam-packed four-day expo and intensive six-day technical program, SC17 has wrapped up another successful event that brought together nearly 13,000 visitors to the Colorado Convention Center in Denver for the larg Read more…

By Tiffany Trader

SC17 Keynote – HPC Powers SKA Efforts to Peer Deep into the Cosmos

November 17, 2017

This week’s SC17 keynote – Life, the Universe and Computing: The Story of the SKA Telescope – was a powerful pitch for the potential of Big Science projects that also showcased the foundational role of high performance computing in modern science. It was also visually stunning. Read more…

By John Russell

How Cities Use HPC at the Edge to Get Smarter

November 17, 2017

Cities are sensoring up, collecting vast troves of data that they’re running through predictive models and using the insights to solve problems that, in some cases, city managers didn’t even know existed. Speaking Read more…

By Doug Black

HPE Extreme Performance Solutions

Harness Scalable Petabyte Storage with HPE Apollo 4510 and HPE StoreEver

As a growing number of connected devices challenges IT departments to rapidly collect, manage, and store troves of data, organizations must adopt a new generation of IT to help them operate quickly and intelligently. Read more…

SC17 Student Cluster Competition Configurations: Fewer Nodes, Way More Accelerators

November 16, 2017

The final configurations for each of the SC17 “Donnybrook in Denver” Student Cluster Competition have been released. Fortunately, each team received their equipment shipments on time and undamaged, so the teams are r Read more…

By Dan Olds

SC Bids Farewell to Denver, Heads to Dallas for 30th

November 17, 2017

After a jam-packed four-day expo and intensive six-day technical program, SC17 has wrapped up another successful event that brought together nearly 13,000 visit Read more…

By Tiffany Trader

SC17 Keynote – HPC Powers SKA Efforts to Peer Deep into the Cosmos

November 17, 2017

This week’s SC17 keynote – Life, the Universe and Computing: The Story of the SKA Telescope – was a powerful pitch for the potential of Big Science projects that also showcased the foundational role of high performance computing in modern science. It was also visually stunning. Read more…

By John Russell

How Cities Use HPC at the Edge to Get Smarter

November 17, 2017

Cities are sensoring up, collecting vast troves of data that they’re running through predictive models and using the insights to solve problems that, in some Read more…

By Doug Black

Student Cluster LINPACK Record Shattered! More LINs Packed Than Ever before!

November 16, 2017

Nanyang Technological University, the pride of Singapore, utterly destroyed the Student Cluster Competition LINPACK record by posting a score of 51.77 TFlop/s a Read more…

By Dan Olds

Hyperion Market Update: ‘Decent’ Growth Led by HPE; AI Transparency a Risk Issue

November 15, 2017

The HPC market update from Hyperion Research (formerly IDC) at the annual SC conference is a business and social “must,” and this year’s presentation at S Read more…

By Doug Black

Nvidia Focuses Its Cloud Containers on HPC Applications

November 14, 2017

Having migrated its top-of-the-line datacenter GPU to the largest cloud vendors, Nvidia is touting its Volta architecture for a range of scientific computing ta Read more…

By George Leopold

HPE Launches ARM-based Apollo System for HPC, AI

November 14, 2017

HPE doubled down on its memory-driven computing vision while expanding its processor portfolio with the announcement yesterday of the company’s first ARM-base Read more…

By Doug Black

OpenACC Shines in Global Climate/Weather Codes

November 14, 2017

OpenACC, the directive-based parallel programming model used mostly for porting codes to GPUs for use on heterogeneous systems, came to SC17 touting impressive Read more…

By John Russell

US Coalesces Plans for First Exascale Supercomputer: Aurora in 2021

September 27, 2017

At the Advanced Scientific Computing Advisory Committee (ASCAC) meeting, in Arlington, Va., yesterday (Sept. 26), it was revealed that the "Aurora" supercompute Read more…

By Tiffany Trader

NERSC Scales Scientific Deep Learning to 15 Petaflops

August 28, 2017

A collaborative effort between Intel, NERSC and Stanford has delivered the first 15-petaflops deep learning software running on HPC platforms and is, according Read more…

By Rob Farber

Oracle Layoffs Reportedly Hit SPARC and Solaris Hard

September 7, 2017

Oracle’s latest layoffs have many wondering if this is the end of the line for the SPARC processor and Solaris OS development. As reported by multiple sources Read more…

By John Russell

AMD Showcases Growing Portfolio of EPYC and Radeon-based Systems at SC17

November 13, 2017

AMD’s charge back into HPC and the datacenter is on full display at SC17. Having launched the EPYC processor line in June along with its MI25 GPU the focus he Read more…

By John Russell

Nvidia Responds to Google TPU Benchmarking

April 10, 2017

Nvidia highlights strengths of its newest GPU silicon in response to Google's report on the performance and energy advantages of its custom tensor processor. Read more…

By Tiffany Trader

Google Releases Deeplearn.js to Further Democratize Machine Learning

August 17, 2017

Spreading the use of machine learning tools is one of the goals of Google’s PAIR (People + AI Research) initiative, which was introduced in early July. Last w Read more…

By John Russell

GlobalFoundries Puts Wind in AMD’s Sails with 12nm FinFET

September 24, 2017

From its annual tech conference last week (Sept. 20), where GlobalFoundries welcomed more than 600 semiconductor professionals (reaching the Santa Clara venue Read more…

By Tiffany Trader

Amazon Debuts New AMD-based GPU Instances for Graphics Acceleration

September 12, 2017

Last week Amazon Web Services (AWS) streaming service, AppStream 2.0, introduced a new GPU instance called Graphics Design intended to accelerate graphics. The Read more…

By John Russell

Leading Solution Providers

EU Funds 20 Million Euro ARM+FPGA Exascale Project

September 7, 2017

At the Barcelona Supercomputer Centre on Wednesday (Sept. 6), 16 partners gathered to launch the EuroEXA project, which invests €20 million over three-and-a-half years into exascale-focused research and development. Led by the Horizon 2020 program, EuroEXA picks up the banner of a triad of partner projects — ExaNeSt, EcoScale and ExaNoDe — building on their work... Read more…

By Tiffany Trader

Delays, Smoke, Records & Markets – A Candid Conversation with Cray CEO Peter Ungaro

October 5, 2017

Earlier this month, Tom Tabor, publisher of HPCwire and I had a very personal conversation with Cray CEO Peter Ungaro. Cray has been on something of a Cinderell Read more…

By Tiffany Trader & Tom Tabor

Reinders: “AVX-512 May Be a Hidden Gem” in Intel Xeon Scalable Processors

June 29, 2017

Imagine if we could use vector processing on something other than just floating point problems.  Today, GPUs and CPUs work tirelessly to accelerate algorithms Read more…

By James Reinders

Cray Moves to Acquire the Seagate ClusterStor Line

July 28, 2017

This week Cray announced that it is picking up Seagate's ClusterStor HPC storage array business for an undisclosed sum. "In short we're effectively transitioning the bulk of the ClusterStor product line to Cray," said CEO Peter Ungaro. Read more…

By Tiffany Trader

Intel Launches Software Tools to Ease FPGA Programming

September 5, 2017

Field Programmable Gate Arrays (FPGAs) have a reputation for being difficult to program, requiring expertise in specialty languages, like Verilog or VHDL. Easin Read more…

By Tiffany Trader

HPC Chips – A Veritable Smorgasbord?

October 10, 2017

For the first time since AMD's ill-fated launch of Bulldozer the answer to the question, 'Which CPU will be in my next HPC system?' doesn't have to be 'Whichever variety of Intel Xeon E5 they are selling when we procure'. Read more…

By Dairsie Latimer

IBM Advances Web-based Quantum Programming

September 5, 2017

IBM Research is pairing its Jupyter-based Data Science Experience notebook environment with its cloud-based quantum computer, IBM Q, in hopes of encouraging a new class of entrepreneurial user to solve intractable problems that even exceed the capabilities of the best AI systems. Read more…

By Alex Woodie

How ‘Knights Mill’ Gets Its Deep Learning Flops

June 22, 2017

Intel, the subject of much speculation regarding the delayed, rewritten or potentially canceled “Aurora” contract (the Argonne Lab part of the CORAL “ Read more…

By Tiffany Trader

  • arrow
  • Click Here for More Headlines
  • arrow
Share This