Swag: Great notebooks from ipHouse, and bright blue T-shirts. Pizza for lunch was OK (although there were still logistics problems--- catering staff running back and forth delivering pizzas a couple boxes at a time to the serving tables.) Crosscut Pale Ale served at the mixer afterwards was very good.
Agile: This impromptu session sort of wandered. What was most interesting was hearing Robert Stephens (founder of Geek Squad) speculate that smart phone adoption might make management more aware of the possibility of quick turnaround or early prototyping. Currently Best Buy outsources a lot of IT development, resulting in a turnaround time of 18 months on major projects.
GPU Computing: Great talk by Paul Cantrell demonstrating some GPU programming on the iPad. A good primer on what the issues and difficulties are. One limitation he encountered which might not be obvious is timeouts--- the OS tended to kill operations that took more than about 5s on the GPU. I had also not been aware that the vertex shader (or other uploaded code) is actually compiled on the GPU. He stated that most of the advice from the chipmaker (of the PowerVR SGX) turned out to be incorrect for his application. For example, they claimed the GPU did loop unrolling; he found that unrolling the loop himself had a 3x improvement.
Hack da Planet: A good talk on productivity tools marred by annoying real-time polls performed via texting. Covered Evernote, Dropbox, 1Password, Dial2Do, Tungle, and TextExpander.
Tao of Systems Architecture: A good premise by Mark Beckman, but not delivered well. He wanted to talk about what characteristics define a "successful" system. Unfortunately, he tended to just read the general principles off his slides, rather than giving specific examples. It was significantly more interesting when he did! There was not a lot of interaction with the audience.
The larger problem, I think, is that he took a very engineering-centric, if not entirely tautological view. A system that matched his principles was "successful." Systems which were operating and thriving in the real world despite not following them were "on life support". Systems which should have been successful but were not were "killed."
But, I would tend to say that we need to take a historical or evolutionary approach as well. Systems (companies, software architectures, procedures) that survive must be doing something right, no matter what flaws we see in them. It is often better to be wrong but first to market rather than right but second (or third. or never.) This means there are tensions among some of the various principles he described, and factors which are very important to an entrepreneur or manager which he tended to neglect.
Lift Bridge Brewing: Fun talk by one of the founders of this (soon-to-be-)Stillwater brewery. All five founders are still working full time elsewhere; the brewery, while thriving, is still somewhat of a hobby. He answered a lot of questions; it was interesting hearing about the logistics of creating your craft beer in somebody else's brewery, and acting as their own distributor. I should have asked what they *hadn't* managed to outsource; he mentionend ProMash, Google Apps, OnePlace, RingCentral, eTrace/Field Force Manager, hopchart.com, and salesforce.com. (HopChart is a beerlog for you to record beers you've tried; plus it makes recommendations based on what people with similar tastes.)
How can MN be Better?: Panel discussion with Dan Grigsby, Chris Smith (Coral Group), Marti Nyman (Altavail Partners and UMN), Jon Dahl (Zencoder, a recent Y Combinator graduate), and Robert Weber (W3i). The discussion focussed mainly on startups, of course. Smith (or Nyman?) spoke approvingly of the recently passed angel tax credit. Much was made of Minnesota's "risk-averse" culture. But I think Dan Grigsby was closer to the mark when he said that there needs to be a population of leaders with proven track records before there will be much equity investment. Without experience in a high-growth company, where will both the people and ideas for "home runs" come from? (He also pointed out that without such companies, there is no good fallback position for failed entrepreneurs.)
Building up this infrastructure/ecosystem is hard work. Nyman told a story about a San Francisco-based VC who said he'd love to see pitches from MN companies--- but, he'd need an investor in town who he trusted (to keep an eye on the company locally) and a team that's had startup experience. Similarly, efforts to build a startup incubator locally have failed; one of the missing factors identified is "fertile ground" for the next round of funding when startups leave the nest.
One thing which has been brought up in previous years (but I don't think got a mention today) that is also part of the ecosystem is a strong research university. This serves to generate both ideas and high-quality workers. But UMN's graduate computer science department is ranked relatively low compared to other centers of innovation. 35th (in the U.S. News rankings for 2010) is not bad, but it's not good either--- Stanford and Berkeley are both first-rank, as is MIT. UW-Madison is ranked 11th in the report.
Stasis: Transactions without SQL got about 15-20 people, and they were mostly engaged and asking questions. It was fun.