Matt Biddulph
Nokia
How does big data connect theories about cities to results and to products we can actually make?
We build our understanding about systems by making models. The problem with models is that to make them work we simplify to falsehood.
All models are wrong but some are useful.
We are at the end of theory. The science of the future will not make progress by making models. Instead of making models, we can now gather data from the real world and analyze the real world.
People are city biology. People make cities too complicated to model. When you look at the movement of bikes around London, it looks like a city breathing.
The street is platform. We can't see all the data that is in the street. We can't measure them. Mobile devices are the city sensors. Phones are packed full of sensors. Phones are no longer blind, they are networked and smart.
The flip side to having all this data is how to process it and make it useful. At some point you can't process all the data you collect with the power you have. Google is trying to tackle this problem on the web. Indexing billions of web pages and data sets.
Map-->Reduce.
Take some data, map it, then reduce it.
The mapping and the reducing stage can be broken up and done in parallel. Turn the problem into something piecemeal and manageable. Add more machines as you have more data, so that it scales.
(Why do we need to process all of the data. Scaling to the size of the world sounds like a fool's errand. Whatever happened to the sample? Processing more data than you need just because you can is idiotic. I'd like to understand in which situations the use of big data is really useful and productive and in which situations it is geek masturbation... Not that I am against such a thing all the time, but I do think it is important to be honest about what we are doing and why.)
Can the phone be our sole source of data? (I don't understand this question. It can be, sure. Should it be? Why should it be?) Does the information translate back into useful and desirable data? Can I get a feel for a city I have never been to because of the data we've sifted and sorted from mobile phones?
We wanted to test the data to see if we can observe patterns and structures that we can show to be true.
For example, we looked at search patterns. What did people look for and where were they when they looked for it? We discovered that people search for Ikea on their phones when they are near Ikea, and we also saw Prenzlauerberger yuppies looking for Ikea from home. Also, we saw a spike on Saturday.
(The real question then is how does this information get back to the user in a way that is beneficial to her and not just helping a corporation to learn how to target advertising.)
When you search you are revealing your goals, but in a mobile phone, we also have a lot more passive context.
The Starbucks index. By processing POI data, we can tell you meaningful things about where you are now. For example, we can tell you how Starbucksy an area is.
If we look at how many times each map tile on ovi.com is loaded, we get an attention map of the world. Which parts of the city do people care about? If we look at patterns from the use of the drive navigation application, we see something else. We see where people who don't know their way around are going and when. Commuter data would be different.
You can use cities you know to check the viability of the data sets so that you can also make predictions about cities you don't know. Validating common-sense predictions.
Design with data... where is the product... how do we make it make sense?
Mike Kuniavsky "Smart Things" is a good book
What happens when the intelligence of the web is embedded in stuff in the real world? Information has a grain and different kinds of data are appropriate for different products. Data as a new raw material (love).
***
Question: I want a company to store my data and give it back to me so I can use it with another service. (Oh, God yes!)
Matt's response, from a personal point of view, not Nokia's. His personal philosophy is that yes, this should be the case. You should be able to get your data out.
(I didn't like it when Matt prefaced that statement by saying that he works at Nokia because his start up was acquired. That may be how he got there, but if that is why he's still there, he should get another job. Anyhow, It was an uncomfortable question and he handled it with aplomb overall.)
