I am excited to share that two new businesses started by good friends that I have been advising have just launched.
iStreamMobile offers branded Internet radio streaming services with innovative content and revenue models.
High Country RawFood is a Colorado-based startup focusing both on creating high quality raw food products (currently available at farmers markets in Breckenridge and Minturn) and on raw food preparation education via classes and through its wHolyFood blog.
A few months ago Curbed SF published an article on 43 Awesome San Francisco Public Parklets. These parklets are a public-private partnership, where the land is publicly owned (usually part of a sidewalk) but a nearby business (usually a cafe) maintains it. The parklet becomes part of the outdoor seating for the business, but anyone can hang out there, even if they are not customers of the business.
I thought it would be interesting to calculate the shortest cycling route that visits all the parklets (the classic Travelling Salesman computer science problem). Since I have been recently working with Mapbox, the rapidly growing alternative to Google Maps et al, I thought it would be a good test case for Mapbox’s relatively new driving directions feature (to calculate the distance of the best route between any pair of parklets, which then goes into the Travelling Salesman algorithm). Unfortunately this only supports vehicle mode right now (no cycling mode) so for now this is the driving version of the shortest route.
The best result I have found so far is 44.46 miles:
I found a Python implementation of a Travelling Salesman algorithm. As the Wikipedia article discusses, this is a surprisingly difficult problem to solve, because the number of permutations grows exponentially as one adds locations.
Hence the route above is not the perfect solution, but merely the best that the algorithm can find in a reasonable amount of time.
I wrote a Python script to pull in the parklets KML, use the Travelling Salesman algorithm together with the Mapbox Driving Directions API to find the best route, and output it to KML.
Just got back from attending the kickoff and final demos of the weekend-long Music Hack Day SF May 2014 held at the headquarters of GitHub in San Francisco. It was tremendously inspirational to be around such focused energy, and the API demos during the kickoff were educational (though very brief).
In the end 41 teams formed, pitched, developed their products in 24 hours, and demoed them. Quality was mixed, though even the unfinished and buggy ones usually had innovative ideas behind them. The APIs used were the usual suspects such as Spotify, RDIO, SoundCloud, EchoNest, GraceNotes and Twitter, with the Leap Motion gesture tracker being a surprisingly popular (and impressive) choice of input device.
The GitHub space was an inspiration in itself, as captured in this recent Wired article on it.
Stir together in a mug:
3 tbsp almond meal (or 2 almond meal plus 1 oatmeal)
1 tbsp whole wheat flour or gluten free flour
1 tbsp cocoa
small pinch baking soda
1 tbsp olive or other oil (or one egg)
approx 2 tbsp milk or coconut milk, to make the consistency like that
of cake batter
Microwave 3 mins until firm on top, then sprinkle choc chips on top.