As noted back in our budget commentary, the Government of Canada seems to have abandoned the idea of a Canadian Advanced Research Projects Agency (CARPA – see here, here and here for previous takes on this) in favour of a yet-to-be-named innovation-and-investment agency said to be based on the Finnish agency Tekes and the Israel Innovation Authority.   A number of people have asked me to explain what the difference is between the two.  As always, I am here to oblige:

The CARPA idea is relatively simple, and the best defense of it you are going to get is in this Public Policy Forum document The Case for a Canadian Research Projects Agency.  Basically, it works like this:

1)            Government creates an arms-length agency and frees it from any imaginable rule that would slow it down – no public sector rules on hiring, procurement, transparency, bilingualism – forget it: CARPA is meant to be lean, mean, and run by a rotating staff numbering in the dozens, not hundreds, and made up entirely of non-career bureaucrats who are not afraid to move fast and break things.

2)            Working in collaboration with government authorities (it’s not entirely clear who that would be, but we’ll leave that aside for now), arms length body comes up with very futuristic “problems” that need solving (current examples from the US’ Defence Advanced research projects Agency include: “Bio-inspired Restoration of Aged Concrete Edifices”, “Modular Efficient Laser Technologies”, and “Adaptive Warfighting Architectures”). Based on calls for proposals, the fast-moving companies, laboratories, etc. receive contracts worth millions.

3)            The point of ARPAs is to be high-risk, high-reward.  So, most of these ideas don’t really pan out.  But a) failure can be a learning experience and b) if your project portfolio is big enough, a few of these projects end up developing technologies that are quite important (so-called “breakthrough technologies.”)

4)            Breakthrough technologies can get picked up, commercialized and – occasionally – turn into big industries in their own right.

The challenges, of course – and what many have found in attempting to clone DARPA – is that none of these four steps is a given.  In Germany, a recent attempt to create an ARPA clone fell foul of the Federal Court of Auditors, which basically said it was not kosher to have an arms length agency working in secret and under an operating framework completely different from the rest of the public sector (something I am fairly sure would also happen in Canada).  The ability to achieve breakthrough technologies using this approach is based on scale.  Even at DARPA, there have probably only been a dozen or so truly breakthrough technologies produced in its 65-year history: all it one every five years.  A proposed CARPA would have had a budget only about a tenth that size: simple math suggests breakthrough technologies would only come around about once every fifty years.  And, last of all, the desire of the private sector to engage in this kind of work and the desire of investors to back the companies that do it depends on the prospect of these technologies turning into future sales and future products.  The US Department of Defence buys a lot of stuff, so companies and investors see a lot of possibilities.  But in other fields it is not so simple, as this Economist article notes. For ARPA-E, the American government’s attempt to replicate DARPA in the energy field, the problem has been that the expected timescale for payback is so long that investors aren’t providing similar funding; and in Canada it is genuinely unclear which federal department(s) would be in a position to take on DoD’s procurement role in any proposed ARPA ecosystem.

Ok, so much for CARPA.  What are these strange Finnish and Israeli models the government is now interested in?  Well, they differ a bit in origin – Tekes (a formerly independent agency that promoted the economy-wide take-up of technology, but is now part of Business Finland) and the Israel Innovation Authority (IIA) used to just be called the Office of the Chief Scientist. They also differ in their actual modes of functioning.  (Click here for a good short explainer on Tekes and click here for an interview on the IIA’s activities). 

Complicating matters a bit further is that both agencies are considered somewhat past their prime – that is they were daring and exciting agencies in the late 1990s/early aughts, but that mainly because they were small and ignored by government and they took advantage of lax oversight to be creative and take risks.  Once they became successful, central governments became more involved and these institutions lost their edge a bit.  So, when you hear innovation nerds talk admiringly about Tekes and the IIA, they are really talking about earlier incarnations of these institutions which were successful in part for some of the same reasons DARPA was – that is, they were science funding agencies that didn’t “act” like governments.

But that’s about where the similarity ends.  Tekes and IIA both have a fairly wide portfolio of activities, but at heart they are both applied research agencies: that is, they exist to funnel public money into projects where businesses have already identified commercial opportunities and want scientific/ technical assistance to bring these things to market.  No moonshots or “breakthrough tech” here – it’s about helping firms achieve their aims by providing them with a mixture of money and talent.  Part angel investor, part Tinder for scientific expertise.  Some programs run by these agencies are very tech-neutral and business-led (i.e. the agency doesn’t care which industry the firm is in and the portfolio is based on whatever business is pitching), while others are very field specific and more directed (the agency picks the industrial field it wants to invest in and then actively uses its funds to bring industry and academia together to work on problems).  It’s not far removed from the German Fraunhofer approach to applied research, except instead of doing direct deals with industry and performing research themselves, Tekes and IIA in effect contract out the research back to consortia of firms and academic institutions.

There are similarities here.  In both cases, you have arms-lengths, preferably very lean organizations staffed by non-career public servants paying researchers (sometimes academics but sometimes not) to do research.  But the purpose of the money and the way it gets used are completely different.  In the CARPA model, you have government/“non-bureaucrats” deciding what the “challenges” are and then paying firms/labs to do some mix of basic and applied science to come up with wild and crazy ideas, under the assumption that many will fail but a few will pay big returns.  In the Tekes/IIA model, you have a slightly less “move-fast-break-things-y” agency working to build capacity in applied science in order to bring new products to market faster.  That means not just feeding money into projects deemed commercializable in the short-term, but in projects and relationships leading to ongoing partnerships that will catalyze more such innovative work in the future.  And if you see that as being complementary to building innovation “clusters” (and not just of the ersatz “supercluster” variety), you’d be right – encouraging cluster formation is one of the things Tekes did very well (at a much lower price tag than our Superclusters).

The last thing I would mention here with respect to Tekes and IIA is that it is not clear how well they would translate into Canada.  As I noted back here, one of the arguments against things like DARPA is that while it makes sense in a big, deeply capitalized market like America’s, it was not clear how it would work in a smaller country like Canada, which does not have the same pools of talent or money.  But one could easily turn that argument around: Tekes and IIA were (at least at the outset) well-suited for countries of 5-6 million people where the vast majority of the population lives within a 3-hour drive from the capital.  But Canada’s a very different beast.  Can this model be made to work here?  Time will tell.