A new Cooperative Research Centre funding round opened this week.
It is for three-year projects to commercialise AI systems and technologies, including robotics and automation.
It might sound really relevant research and while the timing could have been worse, it is hard to see how.
It came as the Strategic Examination of Research and Development proposed closing half a dozen research programmes to fund its grand plan, including CRCs.
SERD simply states that future funding should go to its National Strategic Initiatives, which will “bring together” government, industry and “the research sector to tackle high-risk, high-impact challenges.”
Which is pretty much what CRCs have always done – combining business, universities and public research institutes in time-limited teams with a specific brief to fix a problem or create a product.
CRCs can be cumbersome (too many partners), complicated (too many cultures) and confused (too many objectives). Over the decades the model got manipulated, there were community service CRCs that built constituencies that protected them politically.
But most stuck to the script and closed when their funding ran out, generally after a decade. They either floated or fell into the “valley of death” between lab and marketplace. And they did not leave CSIRO-university style legacy employees who don’t do well when time is tightish and goals are specific. CRCP’s, short term teams to fix a specific problem are faster and cheaper. System expert Tony Peacock says, “it is extraordinary to me that some big-name unis can't manage to average involvement in a single CRC-P per round.”
CRC's have always been selected to address research problems top of policy-make mind. “In many respects (SERD) proposes to create super CRC's – larger and more strategically focused in national priorities,” says Jane O’Dwyer from lobby Cooperative Research Australia.
So why replace the existing model with one where centralised decision makers have a national research plan they control and massive resources to allocate?
The question might be its own answer.