The Missing Translation

Australia’s innovation problem is usually framed as a failure to translate research. In practice, it is more often a failure to translate researchers.

This article was written with the assistance of AI and is part of the Translation Series. Warning - this is a draft written on my phone. I am very keen to publish before the budget, but editing a SubStack article on a phone is a pain.

In March 2026, the Strategic Examination of Research and Development released its final report, Ambitious Australia. Twenty recommendations. A complete redesign of the R&D Tax Incentive. Collaboration vouchers, premium startup streams, points-based eligibility tests, and a new National Innovation Council. It is the most comprehensive review of Australia’s innovation system in a generation, and it is built on the wrong assumption.

The assumption is this: Australia’s innovation problem is a translation problem — specifically, the failure to translate intellectual property from universities into commercial use. Fix the conduits. Build better bridges. Incentivise collaboration. Move the knowledge across the divide.

That diagnosis is not wrong. It is incomplete in a way that makes the prescription ineffective.

The knowledge doesn’t primarily move in documents, patents, or licensing agreements. It moves in people.

The CRC Postdoc and the Demand That Wasn’t There

I know this from the inside.

My postdoctoral research was funded through the Cooperative Research Centre program — one of the genuine strengths of Australia’s innovation infrastructure, and specifically its postdoctoral stream. The CRC postdoc program did what good policy is supposed to do. It trained researchers with bifocal capability: rigorous in method, oriented toward application, exposed to commercial constraints and industrial problems in a way that pure academic training never achieves. By the end of it, I was genuinely prepared to work in industry.

And then there was nowhere to go.

Not because the training had failed. Because the demand side hadn’t been built. Australian firms — the SMEs that constitute the productive base of any innovation economy — were not in the habit of employing researchers. The career infrastructure didn’t exist. The status equivalence wasn’t there. The incentive architecture pointed back toward the public research sector.

So I stayed at CSIRO for the next thirty-six years.

That is not a complaint. CSIRO is where Zebedee was built, where genuine field robotics innovation happened over decades, where the kind of long-cycle research that produces real commercial capability was possible. But as a story about what the Australian innovation system produced with its investment in my training, it is a story about a pipeline that fed back into itself.

The CRC program created the supply. Australia never created the demand.

The Migration Signal

There is a fact about Australian research careers that the SERD report does not confront directly: for many researchers who want to work in industry, the rational career move is to leave the country.

This is not a rumour. It is a structural reality visible to anyone who has watched cohorts of talented researchers navigate their options. The ones who wanted to build careers inside firms — who wanted to do research in the context of commercial constraints, in organisations with skin in the game — largely concluded that Australia didn’t have enough of those organisations, or that those organisations didn’t have enough appetite for research capability. So they went to Germany, the United States, the United Kingdom, the Netherlands. Economies where the researcher inside a firm is unremarkable. Where being a research director at a mid-sized manufacturer carries genuine career status. Where the knowledge doesn’t have to cross a bridge because it was never quarantined on one side of the river in the first place.

Australia subsidised their training. Other countries got the absorptive capacity.

When emigration is the primary pathway for researchers who want industry careers, the problem is not supply. The problem is demand. And demand-side problems are not solved by building better bridges from the supply side.

What SERD Is Actually Doing

Read the SERD recommendations carefully and the orientation is unmistakable. The entire architecture — collaboration premiums, university vouchers, points tests rewarding university partnerships, PhD programs designed to facilitate industry engagement — is built around creating better conduits for knowledge to flow from universities into firms. The university remains the knowledge source. The firm is the recipient. The researcher is the delivery mechanism.

Even the collaboration voucher program — up to $150,000 for SMEs to engage with universities or publicly funded research agencies — reinforces this structure. It doesn’t ask whether the firm has the internal capability to absorb and deploy what the university produces. It connects firms with research institutions. It does not build the demand-side capability that would make those connections productive in the long term.

The RDTI redesign compounds this. An SME that employs a PhD researcher full-time — that has internalised research capability, that has a person on the payroll who reshapes how the firm thinks, who builds enduring technical culture — is largely invisible to the incentive architecture. It may struggle to meet the new $150,000 expenditure floor. It will likely fail the revenue growth test that gates access to the refundable offset. It earns no collaboration premium because its collaboration is internal, not institutional.

Meanwhile, a firm that structures the same researcher relationship as a university contract — keeping them on a university payroll, formalising the engagement, pointing at it in the points test — qualifies. The knowledge transfer is probably worse. The commercial integration is definitely worse. The incentive is better.

A policy that rewards institutional form over substantive capability is not a translation policy. It is a paperwork policy.

The logical endpoint of this architecture is not a more innovative economy. It is a research sector that has learned to behave like a consulting industry — optimised for engagement metrics, structured deliverables, and voucher throughput — and a business sector that has learned to rent knowledge rather than build the capability to generate it. Neither is what Australia needs. And neither is what universities, at their best, actually are.

Researcher Translation

The reframe I want to propose is simple but has significant consequences for policy design.

Australia’s innovation problem is not primarily a research translation problem — the failure to move IP from universities into firms. It is a researcher translation problem — the failure to move researchers themselves into industry, permanently, as a normal career trajectory, at career-forming stages of their professional lives.

IP translation treats knowledge as a product to be transferred. Researcher translation treats knowledge as an embodied capability that compounds when it is embedded in commercial organisations over time. These are not the same problem and they do not have the same solution.

A researcher translation policy would look different from what SERD proposes:

• Incentives attached to employment, not collaboration. An RDTI that rewards firms for directly employing research-qualified people — not for buying episodic access to university expertise — would build enduring absorptive capacity rather than recurring transactions.

• Career recognition that makes industry experience legible. As long as time spent in industry is a career penalty in academic promotion systems, researchers will rationally avoid it. Changing that requires either reforming academic incentive structures as a condition of public funding, or building parallel prestige pathways that make industry research careers genuinely attractive.

• Mobility that is a career move, not a secondment. The National Industry PhD Program is framed as building researchers skilled in university-industry collaboration. That is not the same as building researchers who become industry people. The distinction matters. Collaboration preserves the tether to the university. Translation severs it — or at least makes severing it a viable and respected choice.

• Measurement that follows the person, not the patent. If Australia tracked where researchers work five and ten years after their PhD — not just whether their university filed a patent — the demand-side failure would be visible in the data. At the moment it is largely invisible because the metrics are supply-side metrics.

The Sentence That Hasn’t Changed

The CRC postdoc program trained me to work in industry. Industry wasn’t trained to employ me.

I completed that postdoc in the late 1990s. The SERD report was released in March 2026. In the intervening three decades, Australia has produced many reviews, many recommendations, many redesigns of the RDTI, and a sustained decline in business R&D investment relative to every comparable economy.

What it has not produced is a serious, sustained policy effort to make the researcher in the firm — not the researcher visiting the firm, not the researcher collaborating with the firm from a safe institutional distance, but the researcher *inside* the firm, on the payroll, reshaping its technical culture — the normal, unremarkable, career-positive outcome that it is in the economies that do this well.

Ambitious Australia is an ambitious document. Its diagnosis of Australia’s broken innovation system is largely correct. But its prescription is built around moving knowledge across a divide that better policy would dissolve.

Policymakers are not wrong to worry that many SMEs lack the internal capability to conduct research effectively. But that is precisely why demand-side capability must be built rather than perpetually rented from public institutions.

When you make universities the gateway to innovation incentives, you don’t get more innovation — you get better-branded consulting firms.

Postscript: The Right Name for the Wrong Program

After writing this article, I was reminded of CSIRO’s Switch program — a secondment initiative that places CSIRO researchers inside industry organisations for three months.

Switch is the right name for the wrong program.

The program’s own framing gives it away. Secondments run for three months on the basis that this is sufficient time to develop commercial insights while — and here is the tell — minimising disruption to projects. The researcher goes in, absorbs some industry context, and comes back. The goal is a better-informed CSIRO researcher, not a researcher who has switched careers.

Australia named a program Switch and then designed it to make sure nobody actually switches.

That is not a criticism of the people who built it. A three-month secondment is genuinely valuable. Researchers who have spent time inside firms think differently about their work. But it is a precise illustration of the boundary that Australian innovation policy has never been willing to cross — the line between exposing researchers to industry and actually releasing them into it.

A genuine Switch program would measure success by how many researchers don’t come back. It would treat permanent movement into industry not as disruption to manage, but as the outcome to celebrate. It would be, in the truest sense of the word, a translation program.

We have the name. We just haven’t built what it implies.

Postscript: $1.6 Billion on the Wrong Translation

Australia’s Economic Accelerator is the most direct expression of the IP translation consensus in policy form. Announced in 2022 and funded at $1.6 billion over ten years, it is built entirely around moving research outputs from universities toward commercial application. Funding flows to universities. Industry participates as a partner, a collaborator, an in-kind contributor. The researcher is the knowledge carrier. The university is the institution. The firm is the destination for the output, not the person.

The AEA is not a bad program. It is a coherent expression of a coherent theory — that Australia’s innovation problem is a pipeline problem, and that the pipeline runs from university research through proof-of-concept through commercialisation to market. Build the pipeline well enough, fund it generously enough, and the knowledge will flow.

What the AEA cannot do — by design, not by failure — is build the demand-side capability in firms that would make the pipeline productive at scale. A firm that receives the output of an AEA-funded project still needs someone inside it who can absorb, deploy, and extend what the university produced. If that person doesn’t exist, the pipeline delivers to an empty room.

The AEA funds the translation of research. Nothing in its architecture funds the translation of researchers. A university team that completes an AEA Ignite project and then watches a researcher leave to join the industry partner permanently has, by the program’s own logic, lost a resource rather than achieved an outcome.

That is the tell. When a researcher moves into a firm, the university accounting treats it as a cost. The innovation system should treat it as the point.

$1.6 billion is a serious commitment. It deserves a serious question: how much of it is building lasting absorptive capacity in Australian firms, and how much of it is funding transactions that end when the grant does? Until we ask that question — and measure the answer — we are investing at scale in a theory of change that the evidence does not fully support.