Google Storage and Predictions API

The first part of the evening was a talk by Martin Omander on 2 Google APIs:

• the Storage API

Figure out what you're good at so that you can focus on that (don't build your own telescope, use other people's tools). The Storage API is what Google use themselves, therefore they have people on pager to make sure it stays up so you don't have to. Libraries such as Boto work with it.

• the Predictions API

You feed it training data, over 3 steps: upload, train, predict. It can do categorisation (for e.g. language recognition), or return a number (e.g. real estate valuation). Input data can be text or numbers.

Hackerspaces and formal education

The juicy part of the evening was the panel, despite starting with a very biased intro from the moderator against education in favour of hackerspaces. Asking the panelists to choose a side might have also contributed to limit the discussion.

2 interesting topics straight from the panelists' introductions:

• To work around the lack of decent sysadmins around, John Looney from Google has created a short graduate program, where they train people for 5 months and then hire them.
• James Whelton is a startup founder that is currently working on "Coder Dojo", a project to bring programming and computing to kids. I had a quick chat with James after the panel and they are bringing this to Ireland, starting in Cork: it seems their process is to find mentors in a city, get the group started running events/workshops/classes on a Saturday afternoon and then they move on to the next city. I'm not quite sure it's a sustainable way to build a momentum but I will be following this very very closely, and perhaps mentor when they come to Dublin. Not entirely sure where to get the freshest source of info: Twitter account here, All Ireland (?) blog over there.

Someone mentioned Sugata Mitra's talk on learning without teachers again, I have to make the time to watch this.

A couple of interesting points:

• Hackerspaces as part of formal education, as a society, or part of the University -- someone brought up the problem of evaluation and how to grade work that would come out of such a setting; a panelist remarked that academia has been doing that for decades, with PhDs!
• If going the society way, be careful to build a community. It should help with getting momentum.

Although there was a nice flow when only the panel and moderator were speaking, I was disappointed with how the audience involvement and back and forth with the panel was handled. Some questions were completely ignored after being asked without giving a chance to the panel to answer, and people talking were encouraged to speak faster or less as we were running out of time. I would have preferred less questions explored more fully, rather than a rush of comments and unresolved question marks.

Everybody would agree there is much more to be discussed on the topic :) Thanks to the organisers for setting this up!