October 2010 – Tech Blog

GeoMob has returned after a longer-than-usual hiatus due to other – and definitely very interesting – commitments of our previous Mr GeoMob, Christopher Osborne. It was a very interesting night with the usual format of four presentations covering aspects of research, development and business. Here’s my summary and comments.

Max Howell, @mxcl – Tweetdeck

I’m a bit unsure on how to comment the involvement of TweetDeck into the GeoSocial business.
Max’s presentation has been focused on the integration of their application with FourSquare. It’s a tightly coupled integration allowing users to follow their Twitter friends using the locative power of Foursquare, i.e. putting them on a map. Max gave out some bread for our brains when commenting that “Google Latitude is not good for us because it gives out location continuously, whereas we are looking for discrete placement of users on POIs“: this is a real example of why more-is-not-necessarily-better and, in my opinion, the main reason for which, to date, Latitude has been less successful in catalysing users’ attention on locative services.

However, I’m not totally sure why TweetDeck foresees its future into becoming a platform to integrate Twitter and FourSquare into a single framework. “Other apps put FourSquare functions in a separate window and this is distasteful“. Is it really? And how exactly will TweetDeck benefit, financially but not only, from this integration? “We spent a lot of time on FourSquare integration but unfortunately it’s not much used“. They should ask themselves why.
Their TODO list includes Geofencing which might be interesting so let’s wait and see.

Matthew Watkins, @mazwat – Chromaroma by Mudlark

For those of you who don’t know it yet: Chromaroma is a locative game based on your Oyster card touch-ins and touch-outs. They’re still in closed alpha, but the (not so many?) lucky users (I’ve asked to join the alpha 3-4 times, but they never replied) can connect their Oyster account to the game and take part to some kind of Gowalla for transport, based on the number of journeys, station visited, personal and team targets.

Two things to be considered:
– open data and privacy: upon joining the service, the user account page is scraped for their journeys. Matthew explained they approached TfL to ask for APIs/free access to the journeys data but “due to budget cuts we’re low priority“. Apparently they’ve been allowed to keep on doing scraping. The obvious issue is a matter of trust: why should someone give their oyster account access to a company that, technically, hasn’t signed any agreement with TfL?
This is worrying, as to get journey history data you need to activate Auto Top-up. So you’re basically allowing a third party to access an account connected to automatic payments from your payment card.
Secondly, I can’t understand TfL’s strategy on open data here: if they are not worried about the use Mudlark is doing of such data, why not providing developers with an API to query the very same data? Users’ consent can be embedded in the API, so I’m a bit worried that Chromaroma is actually exposing the lack of strategy by TfL, rather than their availability to work together with developers. I hope I’m wrong.
– monetising: I’m not scared of asking the very same question to any company working on this. What is Mudlark’s monetisation strategy and the business viability of such strategy? It can’t be simply “let’s build travel profiles of participating users and sell them to advertisers” as TfL would have done that already. And if TfL haven’t thought about this, or if they’re letting Mudlark collect such data without even letting them adhere to some basic T&C, we are in serious trouble. However, it’s the declared strategy by Mudlark that does not convince me. Matthew suggests it might be based on target like “get from Warren Street to Kings Cross by 10 am, show your touch-ins and get a free coffee” or on the idea of “sponsor items” you can buy. Does this strategy have a market that is big enough? And, as I’ve already asked, why should a company pay for this kind of advertisement that is potentially available for free? If the game is successful, however, it will be chaos in the Tube – and I’m really looking forward to it 🙂

Oliver O’Brien, @oobr – UCL CASA Researcher

Oliver has recently had his 15 minutes of glory thanks to some amazing live map visualisation of London Barclays Cycle Hire availability. He went further to develop visualisation pages for different bicycle hire schemes all around the world – before he received a Cease&Desist request by one of the companies involved. As a researcher, he provided interesting insight to the GeoMob showing some geo-demographic analysis. For example, weekdays vs weekend usage patterns are different according to the area of the world involved. London is very weekdays-centric, showing that the bicycles are mainly used by commuters. I wonder if this analysis can provide also commercial insight as much as Chromaroma’s intended use of Oyster data.

Thumbs up for the itoworld-esque animation visualizing bike usage in the last 48 hours – stressing that properly done geo-infographic can be extremely useful for problem analysis. Oliver’s future work seems targeted at this, and ideally we’ll hear more about travel patterns and how they affect the usability of bicycle hire schemes. I can’t really understand why he was asked to take some of the maps down.

Eugene Tsyrklevich, @tsyrklevich – Parkopedia

The main lesson of this presentation: stalk your iPhone app users, find them on the web, question them and make them change the negative reviews.
An aggressive strategy that can probably work – and I would actually describe Parkopedia’s strategy as positively aggressive. They managed to get a deal with AA about branding their parking-space-finding-app in exchange for a share of profit.
Eugene’s presentation was more about business management than development. Nonetheless it was incredibly full of insight. Especially on how to be successful when marketing an iPhone application. “Working with known brands gives you credibility, and it opens doors“. The main door that this opened was actually Apple’s interest in featuring their app on the AppStore, leading to an almost immediate 30-fold increase in sales. This leads to further credibility and good sales: “Being featured gets you some momentum you never lose“. This is a good lesson for all our aspiring geo-developers.

The past (and future?) of location

I must say – without making it too emotional – that I feel somewhat attached to geo-events at the BCS as my first contact with the London geo-crowd was there over a year ago, with a GeoMob including a talk by the same Gary Gale who gave a talk last night. That was, at least for him, one company and one whole continent ago – for the rest of us the “agos” include new or matured geo-technologies: Foursquare, Gowalla, Latitude, Facebook and Twitter places, plus our very own London based Rummble, and minus some near-casualties (FireEagle).

Some highlights/thoughts from his talk:

The sad story of early and big players
– early players are not always winners: this can happen in a spectacular way (Dodgeball) or more quietly (Orkut has not technically been a commercial success, for example) – but also
– big players are not always winners: it’s all just a little bit of history repeating, isn’it? Remember the software revolution? The giant IBM didn’t understand it, and a small and agile company called Microsoft became the de-facto monopolist. OS/2 is still remembered as one of the epic fails in software. Remember the Internet revolution? The giant Microsoft had its very own epic fail called Microsoft Network. It took them ages to create a search engine, and in the meantime an agile and young company with a Big G became the search giant. Some years later, the aforementioned Orkut, started by Google as a side project, didn’t have the agility and the motivation to resist to Facebook. The same might happen about location services.

Power to the people
The problem with big players is that they take the quality of data bases for granted. Foursquare et al. found a way to motivate users to keep the POI database constantly updated by using a form of psychological reward. Something that Google hasn’t quite done.

Now monetize, please
Ok, we can motivate users by assigning mayorship and medals. Having a frequently refreshed database is a step ahead. But how do you make money out of it? “Let’s get in touch with the companies and ask for a share of the profit” can work for some brave early adopters. But it will not take long for companies to realize they can use the data – for free – to make business analysis without even contacting foursquare. “Become mayor and get a 10% discount”. What other data analysis should motivate them to pay for it? Knowing where a customer goes next? Where they’ve been before? Maybe to get higher profile in the searches, like in google searches? In the ocean of possibilities, the certainty is that there isn’t yet an idea that works well. “Even Facebook lacks the time to contact the big players to negotiate discounts“. And if you think about the small players it’s even more difficult (but if Monmouth offers me a free espresso I’ll work hard to become their Mayor!).
The way many companies are trying to sell it is still pretty much old economy: sell the check-ins database to a big marketing company, blablabla. Cfr. next point.

Dig out the meaningful data
Ok, we have motivated the users to keep our POIs fresh. But they want to be mayor, so they exploit APIs. Their favourite bar has already a Mayor? They create another instance of the same place. They create their own home. I’ve seen a “my bed”. Is there an algorithmic way to filter out the meaningless data? Surely not in the general case. Moreover, as Gary stressed, simply “selling your database starts eroding its value“. Because the buyer needs to find a use for that mountain of data. As for now, such use is not evident, because most of the data is not meaningful at all.

“If Augmented Reality is Layar, I’m disappointed”
Some time ago I noticed a strange absence of overlap among the geo-crowd and the AR-crowd. The latter presents ideas that have been discussed for years by the former as a “revolution”. One problem is that maybe we have augmented reality but not a realistic augmentation, mostly because of reduced processing power on mobile devices. Ideally you would like to walk down the broadway, see a SuperMario-like green mushroom that gives you an extra shot of espresso (to me it’s like getting an extra-life), catch it, and claim the coffee in the shop around the corner. Unfortunately, GPS is not accurate enough (Galileo might solve this problem soon) and walking down all the time pointing your phone camera to the road will only drain your battery (and probably get you killed before you manage to catch the mushroom). It’s not just an issue of processing power and battery life, though. Even with that, there’s a serious user interaction issue. AR glasses might, partially, solve that, but I can’t really believe that augmenting reality is *just* that and not something that empowers a user’s imagination. Geo-AR is on the boundary between novelty (“oh look, it correctly puts a label on St Paul’s cathedral!“) and utility. And currently on the wrong side of it.

The director’s cut will (not) include recommendations
“I’m sure we’ll make it to the director’s cut” – Alex Housley complained in the typical flamboyant way of the Rummble crowd about being left out of the presentation. “We believe trust networks are the future“. Yes and no. I agree with Alex in the sense that how to provide appropriate recommendations is an interesting research problem (but also here) and the key to monetization of any service. It’s technically not the future, though: Amazon has been using recommendations for years, and I’ve done purchases myself prompted by their recommendations. Trust networks have been extensively used in services like Netflix. What Rummble is trying to do is a more direct way of exploiting trust networks to enrich recommendations, bringing them to the heart of the application. I’m sure that recommendations will play a role in monetizing the geo-thing and that even trust networks may, too. What I’m not sure about is if recommendations will be as they’re now. Without a revolution in the way users perceive local recommendation – that is, a user interaction revolution – they’re not gonna make it. Users need a seamless way of specifying the trust network, and a similarly seamless way of receiving the recommendation.