The Black Hole Economy
Part two of five in the Running In Place series. This series is about why Canada's innovation ecosystem is failing — and what it's going to cost us if nothing changes.
An overview of the series can be found here.
A few years ago, I watched a program with an eight-figure budget and genuine institutional backing do almost nothing it was designed to do.
It wasn’t a lack of effort. The people running it weren’t incompetent. They worked hard, they ran events, they hit their numbers. Participants showed up. Content got delivered. A report was written. By every internal measure, the program succeeded.
And yet the problem it was built to solve — connecting a generation of young, technically capable people to the institutions and industries that needed them — remained almost entirely intact when it was over. The same gaps. The same disconnects. The same people are on the outside looking in.
I’ve seen versions of this story more times than I can count, across municipal economic development offices, provincial agencies, post-secondary institutions, and federally funded programs. The scale changes. The names change, but the outcome remains the same.
This is the first thing you need to understand about Canada’s innovation ecosystem: it is very good at running programs, and very bad at producing change. Those are not the same thing, and we have spent years pretending they are.
What we measure, and what we don’t
Here is how a typical innovation program gets evaluated: How many events did you run? How many participants came? How many dollars did you spend?
These are output metrics. They measure activity. They do not measure whether anything actually changed downstream — whether a young person found a meaningful entry point into the economy, whether a company solved a real problem, whether an institution became more capable of engaging with the community it nominally serves.1 (see appendix: O5 — Blind Spots)
The reason output metrics dominate isn’t stupidity or bad faith. Output metrics are easy to collect, easy to report, and — critically — easy to hit. You can guarantee them in advance. An institution that promises to run 12 workshops and bring in 300 participants can deliver on that promise, regardless of whether those workshops moved the needle on anything. Impact is harder to promise, harder to measure, and much harder to take credit for.
So the system optimizes for what it can control. And what it can control is activity.
The consequence is a funding environment that rewards the appearance of innovation over its occurrence. Dollars flow to programs that produce clean reports, not programs that produce change. And because institutional survival — the budget, the headcount, the mandate — doesn’t actually depend on demonstrating real-world impact, there is no selection pressure for better. The incentive to improve the system is almost entirely absent. What exists in its place is an incentive to maintain it. (see appendix: C16b — Incentive Architecture)
This is not an abstract concern. Canada’s labor productivity growth has been in structural decline for decades, now sitting at roughly one-third of U.S. levels and among the weakest in the OECD.2 Business investment in research and development — the downstream indicator of whether innovation programming is actually producing innovative companies — ranks near the bottom of OECD nations as a share of GDP, and has been falling.3 Canada’s business R&D intensity sits at just 1.1% of GDP — second lowest in the G7 — while the OECD average is 2%.4 Meanwhile, Canada ranks at or near the bottom among G7 countries for patent applications per capita, a rough proxy for whether publicly funded research is translating into anything deployable.5 We are spending on the infrastructure of innovation and not getting innovation. That gap has a structural explanation, and it lives in the incentive architecture described above.
The internal problem nobody talks about
It would be one thing if different organizations were simply working from different playbooks. You could at least map the territory, identify the gaps, and build something to bridge them.
The more corrosive reality is that individual organizations frequently lack a coherent internal strategy — let alone one that connects to the actors sitting in adjacent buildings. A provincial agency’s workforce development team and its entrepreneurship team may be pursuing the same demographic, deploying similar programming, and have never sat in the same room to discuss it. A post-secondary institution’s research office, technology transfer office, and student entrepreneurship program are on the same campus and share almost nothing else.
There is no cohesive plan. Not across the ecosystem. Often not within a single organization.6
This matters because it means the fragmentation isn’t a coordination problem that can be solved with a cross-sector table. It’s built into the architecture of individual institutions. Adding more tables — more summits, more convenings, more steering committees — doesn’t fix internal incoherence. It papers over it while burning the calendar of everyone involved.
What actually gets produced in the absence of a coherent strategy is programming that reflects internal rather than external political realities. The problems the ecosystem claims to be solving are not the problems it is actually organized to address. And because the reporting doesn’t require honest accounting of that gap, nobody is formally required to confront it.
Canada’s federal advisory bodies have already flagged this. The Council of Canadian Academies’ 2025 report documents that Canada excels in research and talent but faces a “persistent, well-documented failure to translate research into commercial or community outcomes,” pointing to fragmentation and weak university-industry connectivity as primary mechanisms.7 The Canadian Science Policy Centre’s 2025 panel on governing innovation likewise concluded that Canada’s innovation system suffers from “weak business dynamics, limited competitive pressures, and misaligned incentives” that prioritize inputs over long-term socio-economic outcomes. That report is from 2025.8 The diagnosis has been consistent for years. The ecosystem has not meaningfully changed.
Where the money actually goes
I am not describing a funding shortage. Canada spends real money on innovation, on ecosystem development, on talent pipelines, on the infrastructure of opportunity. Federal and provincial governments collectively deploy billions annually through granting councils, innovation agencies, and ecosystem development programs.9 The problem is not the volume of investment. The problem is what that investment is actually buying.
A significant share goes to sustaining the overhead of institutions that are, at best, running in place. Senior leadership positions in ecosystem organizations drawing six-figure salaries, overseeing programming that produces outputs but not outcomes, in service of a mandate that doesn’t require them to demonstrate the difference. (see appendix: O21b — Gaming Vector) That money is not nothing. Redistributed effectively — not in a lump sum, but as targeted, right-sized support to the grassroots organizations doing actual community building on actual budgets — it would produce a fundamentally different ecosystem. (see appendix: C8 — Microgrants)
I want to be precise here because it’s easy to misread this as an argument for defunding institutions and flooding grassroots groups with cash. It’s not. Overfunding grassroots organizations is its own failure mode — it disconnects them from the community logic and intrinsic motivation that made them effective in the first place. The argument is about proportion, and about what kind of infrastructure you build with what you have.
The argument is that we are currently spending a lot to maintain systems that are not working, and very little to build the connective tissue that would allow the things that are working to reach scale.
The honest broker problem
There is one more piece of this that doesn’t get named enough.
Even when you can get to the table — and for most grassroots organizations, most of the time, getting to the table is itself an exercise in attrition — you frequently don’t have an honest broker on the other side. The person across from you works for an institution whose continued existence doesn’t depend on telling you what’s actually true. They have every incentive to tell you what fits within their mandate, what their program can accommodate, and what their reporting cycle requires. They may be a perfectly good person doing a structurally constrained job. But they are not in a position to give you an honest accounting of whether the system is actually working.
This is not merely cynicism, but rather a description of how institutional incentives operate. And it means that the feedback loop required to actually improve the system — honest, ground-level information flowing back to the people with the resources and authority to change things — is almost entirely absent. The people who know what’s broken are not in the rooms where decisions are made. The people in those rooms are not required to seek them out. (see appendix: A25 — Adaptive Measurement)
What this produces
A system that consumes resources and reports success while the underlying problems it was designed to address continue to compound.
Canada’s productivity gap is not a mystery. The problem is a persistent, well-documented failure to translate research into commercial or community outcomes — a finding that is consistent across nearly every serious review of Canadian innovation policy conducted in the last ten years.10 The reviews keep arriving. The model keeps running. (see appendix: A21 — Immune Function)
Producing an ecosystem that looks functional from the inside of an institution and looks like a closed door from everywhere else.
A generation of people — technically capable, genuinely motivated, in their prime — who have spent years trying to plug into systems that are not actually organized to receive them, and who are drawing increasingly rational conclusions about whether it’s worth continuing to try.
That last part is where we’re headed next.
Eden Redman is the CEO and Executive Director of the Network for Applied Technology, a builder-focused innovation community connecting academia, industry, and entrepreneurship across Alberta and Canada. He has been pushing on the same stubborn problems in Canada’s innovation ecosystem long enough to know exactly where the walls are — and hasn’t stopped.
The Network for Applied Technology (NAT) is a talent-based innovation community. It is dedicated to applied technology across a growing range of high-impact domains. NAT promotes education, skill development, access to specialized tools and hardware, and community. The network connects people across academia, industry, and the entrepreneurial sector. Its goal: drive interdisciplinary, scalable innovation across Alberta and Canada.
Next: Roots Without Soil — on the community builders doing extraordinary work on near-zero resources, and what it would actually take to support them.
Appendix
Each article in this series names a structural problem and points toward the same structural answer: an architecture that follows directly from the diagnosis. That architecture is the Ecosystem Health Tracker (EHT), being developed by the Network for Applied Technologies (NAT) — a decentralized coordination layer, not a program or a database. Infrastructure: shared scaffolding, signal architecture, and neutral brokerage that let the things already working in Canada’s innovation ecosystem find each other, interface with institutions, and compound rather than dissipate.
Ecosystem Health Tracker (EHT) — entries relevant to this article
The entries below are from the EHT design log. As of this series the log contains 35 entries across three levels: 14 observations, 12 commitments, and 9 architectural commitments. Every entry traces upward to at least one architectural commitment. The cross-references in superscript brackets point to the DPUG post in which each entry first appeared.
O5 — Blind spots ◇ Observation
Formation: Measurement systems that rely on institutional reports about the ground are one step removed from what they claim to measure — and the distance compounds. Over time the system optimizes for the reports, not the underlying outcomes. The blind spot is structural, not accidental.
Implication: The coordination layer must fund ground-level synthesis as a first-order activity, not a reporting overhead. It must be able to update its own model from direct evidence rather than from the output of the system it is supposed to be measuring.
O21b — Gaming vector ◇ Observation
Formation: Every measurable target becomes a gaming target. This is not a speculative risk; it is the observed behaviour of every grant architecture and measurement system that has operated at scale in this space. Actors optimize for the metric rather than the underlying goal, and the metric stops corresponding to the thing it was supposed to measure.
Implication: A coordination layer cannot rely on the unassailability of its metrics. It must assume they will be gamed, treat the attempt as a first-order signal, and have a live layer empowered to update the measurement in response. The immune function is less a fraud detection bolted onto a static metric; more a structural commitment that metrics themselves are provisional.
C8 — Microgrants ◈ Commitment
Formation: The current funding architecture concentrates overhead at the institutional level and starves the grassroots. Direct-to-builder microgrants at the $100–500 scale, contingent on participation in shared EHT processes, bypass that concentration. The activation is the signal; the signal is the activation.
User: grassroots builders, ecosystem operators running the intake layer, funders seeking direct-to-builder deployment.
Interaction: microgrants are disbursed on demonstrated participation in coordinated EHT activities — roundtables, sprint activations, signal-layer contributions. No institutional intermediary, no administrative overhead capture, no reporting burden the builder has to translate their work into.
Value back: resources reach the people doing the work, at the velocity the work requires. The funder receives a signal tied to actual activation rather than to the overhead layer that currently absorbs most of the flow.
C16b — Incentive architecture ◈ Commitment
Formation: The incentive structure is the design. A coordination layer that rewards legible outputs over substrate building will produce legible outputs over substrate building — regardless of what its stated mission is. Incentive architecture must be designed so that the things that get rewarded are the things the ecosystem actually needs.
User: funders, institutional partners, every participant whose behaviour is shaped by what the layer rewards.
Interaction: the incentive architecture is explicit, auditable, and periodically re-examined. What gets rewarded is a first-order design decision, not a downstream consequence of what was convenient to measure.
Value back: the layer stays pointed at the outcomes it is designed to produce, rather than drifting toward whatever outputs are easiest to count. The architecture holds its shape against the pressure to reward activity over impact.
A21 — Immune function ◆ Architecture
Formation: Capture is not a bug to be patched; it is a steady-state pressure. Every grant architecture, measurement system, and coordination layer ever built at scale has faced it. An immune function is a structural commitment that the architecture will surface gaming attempts quickly, that operators will be empowered to act on what they surface, and that the measurement layer itself can be updated in response rather than defended.
User: the full ecosystem — but especially operators, institutional auditors, and funders whose signal depends on the layer’s honesty.
Interaction: the immune function runs as a continuous process, not a periodic audit. Operators flag anomalies into the design log, adaptive measurement surfaces drift between the signal and ground-level activity, and the phase operating model allows the architecture to update its own entry criteria in response. Capture attempts become data about what the next iteration of the layer needs to be resistant to.
Value back: the coordination layer is not one capture event away from collapse. It is designed to absorb capture attempts, learn from them, and come out of the event with the immune function measurably stronger.
A25 — Adaptive measurement ◆ Architecture
Formation: Any measurement system deployed at scale will drift. The map diverges from the space. The architecture must be designed so that the measurement layer itself can be updated in response to the ground rather than defended against it — the feedback loop that keeps the coordination layer honest over long timescales.
User: institutional funders, operators maintaining the layer, every participant whose interaction with EHT is mediated through measurement.
Interaction: adaptive measurement is continuous, not periodic. Signal-layer intake surfaces drift; operators escalate; the design log records the update; subsequent measurement incorporates the revision. The measurement is provisional; the commitment to updating it is not.
Value back: the layer remains pointed at the actual ecosystem rather than at its model of the ecosystem. Drift becomes a detectable, correctable condition rather than a structural failure mode that accumulates invisibly until the measurement is unrecoverable.
Edges load-bearing for this article’s argument
O5 → C7 → A25 (causal traceback). The blind-spots observation grounds the signal-layer commitment (C7, introduced in Article 1’s appendix), which traces upward to adaptive measurement. Honest signal at intake is the precondition for adaptive measurement being possible.
O21b → A21 (causal traceback). The gaming-vector observation grounds the immune function directly. The immune function does not exist as a response to failed systems; it is built in because the pressure is steady-state.
C8 → A17 (causal traceback). Microgrants become the activation entry of Phase I once the phase operating model is named. The commitment matures structurally — no longer a standalone entry in the log.
References
Canadian Science Policy Centre (CSPC). (2025). Governing Innovation: Designing Institutions for Canada’s Long-Term Prosperity. Panel findings on outcome metrics, institutional incentives, and the gap between output-based evaluation and socio-economic impact.
🔗 https://sciencepolicy.ca/posts/114-governing-innovation-designing-institutions-for-canadas-long-term-prosperity/
Organization for Economic Co-operation and Development (OECD). (2025). OECD Economic Surveys: Canada 2025
🔗 https://www.oecd.org/en/publications/2025/05/oecd-economic-surveys-canada-2025_ee18a269/full-report/executive-summary_897fc08f.html
OECD. (2025). OECD Economic Surveys: Canada 2025. Business R&D spending as percentage of GDP (1.1% of GDP, compared to OECD average of 2.7%).
🔗 https://www.oecd.org/en/publications/oecd-economic-surveys-canada-2025_28f9e02c-en.html
U15 Canada. (2025, October 27). Opening Remarks to the Standing Committee on Science and Research. (Business R&D intensity: 1.1% of GDP, second lowest in G7.)
🔗 https://u15.ca/publications/statements-releases/opening-remarks-to-the-standing-committee-on-science-and-research-october-27-2025/
Council of Canadian Academies (CCA). (2025). The State of Science, Technology, and Innovation in Canada 2025. (Primary finding: "Canada faces a worsening productivity crisis... a persistent failure to translate research into commercial or community outcomes.")
🔗 https://cca-reports.ca/canadas-innovation-performance-continues-to-decline-at-a-time-of-unprecedented-change/
Council of Canadian Academies (CCA). (2025). The State of Science, Technology, and Innovation in Canada 2025. (Fragmentation, weak university-industry connectivity, and misaligned incentives across institutions.)
🔗 https://cca-reports.ca/canadas-innovation-performance-continues-to-decline-at-a-time-of-unprecedented-change/
Council of Canadian Academies (CCA). (2025). The State of Science, Technology, and Innovation in Canada 2025. (Primary finding: “Canada faces a worsening productivity crisis... a persistent failure to translate research into commercial or community outcomes.”)
🔗 https://cca-reports.ca/canadas-innovation-performance-continues-to-decline-at-a-time-of-unprecedented-change/
Canadian Science Policy Centre (CSPC). (2025). Governing Innovation: Designing Institutions for Canada’s Long-Term Prosperity. Panel findings on outcome metrics, institutional incentives, and the gap between output-based evaluation and socio-economic impact.
🔗 https://sciencepolicy.ca/posts/114-governing-innovation-designing-institutions-for-canadas-long-term-prosperity/
Statistics Canada. (2025, November 27). Gross domestic expenditures on research and development (GERD), 2023 (final), 2024 (preliminary) and 2025 (intentions). The Daily. (Total R&D spending: $57.4 billion in 2023.)
🔗 https://www150.statcan.gc.ca/n1/daily-quotidien/251127/dq251127g-eng.htm
Council of Canadian Academies (CCA). (2025). The State of Science, Technology, and Innovation in Canada 2025. (Primary finding: "Canada faces a worsening productivity crisis... a persistent failure to translate research into commercial or community outcomes.")
🔗 https://cca-reports.ca/canadas-innovation-performance-continues-to-decline-at-a-time-of-unprecedented-change/