Take a look out of the window. How would you rate the view, on a scale of one to ten?
Your response can probably tell us a little about the beauty, or otherwise, of the area around you. That’s the premise that ScenicOrNot, one of the mySociety sites that we recently stopped running, was founded on. Happily, ScenicOrNot has now found a home and will continue under new ownership: more about that in a future blog post. Meanwhile, we’d like to celebrate it with a potted history.
An exercise in crowdsourcing, ScenicOrNot served up a series of random images, each representing one square kilometre of Great Britain, and invited users to rate them (the images were sourced from the Geograph project, itself a fascinating open source repository). The results fed into a database of ‘scenicness’.
ScenicOrNot collected that data and also permits anyone to download it, under an Open Data Licence, for their own ends.
What was it for?
To understand why we made ScenicOrNot, you have to go back to the beginnings of our transit-time mapping technology, Mapumental.
Mapumental shows journeys in terms of how long they take, and it was intended to help people make decisions about where to live, work, or go on holiday. We’d figured out how to display bands of public transport journey times, but we knew that those weren’t the only factors that feed into such important life choices.
House prices, average salaries, and, yes, the beauty of the surrounding area all have a part to play. We wanted to be able to add them to Mapumental so that users could get a really rounded picture.
But while there are public databases for house prices and average salaries available, until the creation of ScenicOrNot, there was no such thing for scenicness. There was just one solution: we would have to make our own.
‘Hot or Not’ for scenery
Rather than go and look at every part of the country ourselves, it was time to harness the wisdom of the crowd.
ScenicOrNot, the building of which was managed by the Dextrous Web, launched in 2009. It served users with an unending random series of images showing landscapes from around the country, and was an early foray into both crowdsourcing and gamification for mySociety.
Rating images, as also seen in Kittenwar (and other, less fluffball-centric sites like HotOrNot) is a pleasingly compulsive activity, and within just a few months, every kilometre of the country had been rated at least once.
And as time went by, we reached a critical milestone: the project amassed a minimum of three votes for each image, helping to ensure that the results were less likely to be skewed by eccentric or unusual opinions about what makes a place scenic.
Slotting ScenicOrNot into Mapumental
We now had our ‘scenicness’ data, and house price and salary data from other sources. The decision we made about how to incorporate these data sets was an important one which has worked well for subsequent Mapumental projects like the work we did for the Welsh government, or for the Fire Protection Association.
Effectively, you can think of each data set as a map layer, which may be slotted in our out, as needed. Our showcase site Mapumental Property demonstrates this – it’s effectively the vanilla transit-time Mapumental, with a house price layer (from Zoopla) added in.
A new lease of life
If we hadn’t found a new owner for ScenicOrNot, we’d have shut it down. Happily, though, it’s found a new home and a whole new purpose: we’ll be explaining more about that in our next blog post.
In January last year, at our yearly staff and volunteers retreat, we decided that TheyWorkForYou should do something special for the general election. We decided that we wanted to gather information on where every candidate in every seat stood on what most people would think were the biggest issues, not just nationally but locally too.
Our reasons for setting this ambitious goal were two fold. First, we thought that pinning people down to a survey that didn’t reward rhetorical flourishes would help the electorate cut through the spin that accompanies all elections. But even more important was to increase our ability to hold new MPs to account: we want users of TheyWorkForYou in the future to be able to see how Parliamentary voting records align with campaign statements.
This meant doing quite a lot of quite difficult things:
- Working out who all the candidates are (thousands of them)
- Working out how to contact them.
- Gathering thousands of local issues from every corner of the country, and quality assuring them.
- Developing a balanced set of national issues.
- Sending the candidates surveys, and chasing them up.
The Volunteer Army
This has turned out to be a massive operation, requiring the creation of the independent Democracy Club set up by the amazing new volunteers Seb Bacon and Tim Green, and an entire candidate database site YourNextMP, built by another new volunteer Edmund von der Burg. Eventually we managed to get at least one local issue in over 80% of constituencies, aided by nearly 6000 new volunteers spread from Lands End to John O’Groats. There’s at least one volunteer in every constituency in Great Britain, and in all but three in Northern Ireland. Volunteers have done more than just submit issues, they’ve played our duck house game to help gather thousands of email addresses, phone numbers, and postal addresses.
What we ended up with is a candidate survey that is different for every constituency – 650 different surveys, in short. The survey always contains the same 15 national issues (chosen by a politically balanced panel held at the Institute for Government) and then anything between zero and ten local issues. We’ve seen everything from cockle protection to subsidies for ferries raised – over 3000 local issues were submitted, before being painstakingly moderated, twice, by uber-volunteers checking for for spelling, grammar, obvious bias and straightforward interestingness (it isn’t really worth asking candidates if they are in favour of Good Things and against Bad Things).
In the last couple of days we’ve started to send out the first surveys – we’ve just passed 1000 emails, and there are at least 2000 still to be sent.
We’re aiming to release the data we are gathering on candidates positions on 30th April. We’ll build a nice interface to explore it, but we also hope that others will do something with what we are expecting to be quite a valuable dataset.
Candidates are busy people, so how do we get their attention? Happily, some candidates are choosing to answer the survey just because TheyWorkForYou has a well know brand in the political world, but this has limits.
The answer is that we are going to ask Democracy Club, and it’s army of volunteers to help. We’ll shortly roll out a tool that will tell volunteers which of their candidates haven’t taken the opportunity to go on the record , and provide a range of ways for them to push for their candidates to fill it in.
It would be a lie to say we’re confident we’ll get every last candidate. But we are confident we can make sure that no candidate can claim they didn’t see, or didn’t know it was important to their constituents. And every extra voice we have makes that more likely.
Matthew’s just updated ScenicOrNot, the little game that we built to provide a ‘Scenicness’ dataset for Mapumental, to include a data dump of the raw data. The dump will update automatically on a weekly basis, but currently it contains averaged scores for 181,188 1*1km grid squares, representing 83% of the Geograph dataset we were using, or 74% of all the grid squares in Great Britain. It is, in other words, really pretty good, and, I think, unprecedented in coverage as a piece of crowd sourced geodata about a whole country.
It’s available under the Creative Commons Attribution Noncommercial 3 Licence, and we greatly look forward to seeing what people do with it.
The goal of ScenicOrNot is to be a gentle-ambling sort of quasi game that’s just compelling enough to keep clicking on, just in case the next picture is the dream valley in which you wish to be buried.
mySociety’s obviously not in the business of building games for their own sake, though. This is another crowdsourcing experiment to solve a specific problem – we need a scenicness map of the UK for a major upcoming mySociety project, and there ain’t one to be had any other way, for love or money.
So if you like mySociety, or just want to ogle the best and worst of this Island, please have a play.
There’s been a lot written recently about the cognitive surplus, a phrase coined by Clay Shirky to describe the amount of human energy that can be deployed to create things if only barriers are lowered and incentives sharpened.
mySociety has recently been fortunate enough to see a little of this phenomenon through the explosion of volunteering activity which grew up around our TheyWorkForYou video timestamping ‘game’. For those of you not familiar, we needed video clips of politician’s speaking matched with the text of their speeches, and in just a couple of months a gang of volunteers new and old have done almost all of the video in the archive. Other, much larger examples include reCAPTCHA and the ESP game.
Reflecting on this, my friend Tom Lynn suggested that there was a gap in the market for a service that would draw together different crowdsourcing games, ensure that their usability standards and social benefit were high, and which then syndicate them out in little widgets, recaptcha style, to anyone who wanted to include one on a web page.
This is where Mozilla and Ubuntu come in. Anyone who uses Firefox knows what the home page is like, essentially the Google homepage with some Firefox branding. Ubuntu’s default browser homepage, post patch upgrade especially, is similarly minimalist and focused on telling you what’s changed.
Therein lies the opportunity – using pieces of these default home pages (maintained by organisations that claim to have a social purpose, remember) for more good than simply repeatedly reminding users about the the brand of the product. Traditionally that would mean asking people to donate or become volunteers, but the new universe of ultra-easy crowdsourcing games are challenging that assumption.
Here’s a scenario. One time in ten when I load Firefox, the homepage contains a widget right under the search box that contains an almost entirely self explanatory task that contributed to the public good in some way. This could be spotting an object on a fragment of satellite photo after a disaster, typing in a word that’s difficult to OCR, timestamping a video clip, or adding tags to an image or a paragraph of text. The widgets would be syndicated from the central repository of Cognitive Surplus Foundation ‘games’, and would help groups like Mozilla and Ubuntu to show themselves to millions of tech-disinterested users to be the true 21st century social enterprises that they want to be.