The limits of openness

by Glyn Moody

The rise of open source is now being repeated by open endeavours in other fields, following a fairly constant pattern. First, somebody starts a small, personal project, often almost accidentally, and without any long-term plans. Crucially, they share that project online, and other people join in. Then, the project starts to grow and become quite useful. Later, it begins to rival commercial offerings, and companies start to attack it. Finally, it equals then surpasses those commercial offerings, and the companies find themselves in trouble.

GNU/Linux is the first and best-known example of this progression. Then it happened with Wikipedia: remember this comment from a former Editor-in-Chief of Encyclopaedia Britannica?

The user who visits Wikipedia to learn about some subject, to confirm some matter of fact, is rather in the position of a visitor to a public restroom. It may be obviously dirty, so that he knows to exercise great care, or it may seem fairly clean, so that he may be lulled into a false sense of security. What he certainly does not know is who has used the facilities before him.

That was back in 2004. More recently, this happened:

Today we’ve announced that we will discontinue the 32-volume printed edition of the Encyclopaedia Britannica when our current inventory is gone.

Now we are seeing the same cycle for OpenStreetMap (OSM). It's reached the stage where major users are jettisoning commercial offerings like Google Maps, and adopting OSM. Meanwhile, the attacks have begun:

Despite the positives, recent studies have highlighted some major drawbacks of open source mapping, specifically with regard to safety, accuracy and reliability. In one particular instance, a leading open source map was compared against a professional TomTom map, and shown to have a third less residential road coverage and 16% less basic map attributes such as street names. Worse still, it blended pedestrian and car map geometry, and included ‘a high number of fields and forest trails’ classified as roads.

Indeed the major benefit – the community aspect – has itself presented problems, leaving maps wide open to attack. A highly-publicised case saw a leading provider suffer over 100,000 individual attacks, including reversals of the recorded directions on one-way streets.

The histories of GNU/Linux and Wikipedia show us how this will end, but that doesn't mean mapping companies will meekly go without a fight. In an attempt to flee competition, they're moving upmarket – literally, by adding the dimension of height to their maps:

Since 2006, we’ve had textured 3D buildings in Google Earth, and today we are excited to announce that we will begin adding 3D models to entire metropolitan areas to Google Earth on mobile devices. This is possible thanks to a combination of our new imagery rendering techniques and computer vision that let us automatically create 3D cityscapes, complete with buildings, terrain and even landscaping, from 45-degree aerial imagery. By the end of the year we aim to have 3D coverage for metropolitan areas with a combined population of 300 million people.

It's not just Google adopting this approach: it seems that Apple, too, will be taking to the skies.

One reason why OSM has been able to match services from multi-billion dollar companies like Google and TomTom is that it could modularise the problem and its solution by deploying huge numbers of volunteers on the ground, and then combining their data. That's the classic distributed approach that works so well for both GNU/Linux and Wikipedia. But is it possible to apply the same approach to 3D mapping, and launch thousands of small-scale aerial surveys to keep up with Google and Apple?

Maybe it is. By a happy coincidence, the field of open source drones/robot planes is developing rapidly. For example, a project called Open Relief has just been launched:

We are creating a robot plane to investigate and map disaster zones. This robot is small enough to be launched from footpaths and smart enough to recognise roads, people and smoke. It uses sensors to measure weather and radiation. The information it collects can easily be shared with disaster management systems like Sahana Eden.

Sahana Eden is open source, as is all the software that will be used to control the Open Relief plane.

Another open drone project is OpenPilot.org:

OpenPilot is an Open Source autopilot platform for small UAVs. It's capable of flying multirotors, helicopters and fixed wing aircraft.

OpenPilot is a non-profit project that aims to make UAV technology more affordable for research purposes and humanitarian causes whilst having a lot of fun in the process.

Although not specifically about open source drones, the site DIY Drones has news and resources that cover this rapidly-advancing area.

So that seems good news: for a modest outlay it is possible to buy and operate an open drone that could easily be used to carry out the kind of aerial mapping that Google and Apple are planning, albeit on a much smaller scale. Moreover, the fit between open source, open maps and open drones is quite natural, since you need the first and last of these to update the second of them, which can then be used by open source software and drones in a neatly iterative process.

But if people start making such low-cost drones and deploying them widely in order to, say, enhance OSM with 3D maps, this would take openness into increasingly public realms. Already there are signs of a backlash against Google's and Apple's plans, with people disturbed at the idea of pictures being taken of them from the sky without their knowledge or agreement.

For Google, that's an extension of the problems it met with Street View, and so it probably has mechanisms in place to blur faces and remove details that might expose people to risks (for example by revealing information about household security – or lack of it). Apple, too, will simply spend money to address these issues. But how can volunteer projects like OpenStreetMap – or some new open source drone mapping project – handle this?

It's not at all clear. That's because openness has hitherto been something that geeks have done in the privacy of their bedrooms or classrooms, not publicly. It's only recently, as the open data movement has gained momentum, that issues of privacy have started to be raised. The problem there is that apparently anonymous personal data often isn't, and can be combined with other datasets to reveal the identity of the people concerned. That's one of the reasons why governments are rightly chary about releasing any kind of personal data at all: however anonymous it might seem now, it could be that later releases of unrelated information will allow all kinds of unexpected – and unwanted – correlations to be revealed.

It is this, rather than any technical constraints, that could prevent the flourishing of advanced open mapping projects using open source drones. And it won't just be here, either. Other fields where openness is likely to be stymied by privacy considerations include open genomics – making anonymous DNA sequences freely available – and open data associated with the Internet of Things. If we want openness to thrive in these important new areas, we need to start thinking now about how to address the important privacy issues they raise.

Follow me @glynmoody on Twitter or identi.ca, and on Google+. For other feature articles by Glyn Moody, please see the archive.