As you've rightly focused on, connection (at a personal and business level) is what drives innovation and company formation (see Proximity Geography for studies on social and physical connection driving innovation). That is partly a function of creating opportunity to connect and partly a function of the diversity in the interests and businesses in the connection pool. And that diversity is largely driven by the number of people in the pool.

Silicon Valley treats the world is its (diversity) hinterland. The emerging Asian centers use their size to similar effect. What both have is an incredible local pool of talent tied to International reach for ideas and trends.

I think to reach the scale needed to compete we need to develop Toronto/Ottawa/Montreal into single community. This requires us to solve a number of issues like business structure, work flow, accountability and work and non-work collaboration that are likely to become critical to a global economy increasingly driven by specialization. (Assuming you believe as I do that vertically integrated companies will become less competitive as the global economy evolves)

The other advantage is that each city has different core skills and business composition increasing the opportunities for innovation at the boundary of two disciplines.

On the government front we need to an acknowledgement that businesses fail & and rather than managing to prevent failure (by supporting safe bets and uncontroversial programs) they need to embrace failure as a part of the development and learning process. We also need to celebrate successes, in part by acknowledging the failures that gave rise to it.

In terms of helping the development of the T/O/M triangle government could support high speed video conferencing to allow events like DemoCamp run in each city to be 'attended' by parties in the other cities. This allows people to get a sense of what's going on in other cities and identify people and businesses that are mesh with theirs. To augment this programs to support a culture of collaboration & with models and training on working together while protect ones interests, etc.

Another is supporting incubation through co-working. Fixing the roadblocks in licensing university & government research is another, and grants (hey if they did video links why not do a ResearchCamp & where researchers could discuss their work and its potential applications)

Oh!! I've gone on enough & but I can't leave without chiding Ottawa's OCRI, which challenged by similar questions you raised in your post had decided that the program we need is to fix things is & get this & a marketing campaign to tell us the glass is really half full & not half empty.

http://www.ragobeer.com/strategy_to_win/2008/02/torontos-techno.html

Seems clear enough to me.. if we were Michigan.

2. Effective working relationships between U of T and 416/905 companies that aren't already worth a billion dollars. (Profs don't come to *camp; startups don't engage with the PEY program– which admittedly still appears not to know about HTTP redirects. sigh)

3. Someone on city council (or at Queen's Park) to tell the powers that be how Silicon Valleys really work. (I nominate David; I'll even pay for a barber to make him presentable.)

4. A unicorn.

Odds are about equal on all four.