The following is a set of questions which Bhaveen Dattani put to me, as part of his studies of VGI and OpenStreetMap for his course at Aston university. The basic questions are the always the big questions, and I had to take a step back and think a bit about all the broad issues around OpenStreetMap (my big hobby). In the spirit of openness I’m sharing these answers here:
What is Volunteered Geographic Information (VGI)? / Have you seen VGI?
I have noticed the term VGI used extensively in academia. There are several terms used for the same concept. Technologists will refer to the same (or similar) concept as “Crowd-sourced” geographic information.
But in fact, when describing the project I am involved in, OpenStreetMap, I prefer the term “mass collaboration”. Some VGI initiatives are mostly about “sourcing” data on the cheap from a crowd of low-skilled contributors. You can think of OpenStreetMap in those terms, but OpenStreetMap has volunteers who bring a wide range of skillsets and levels of dedication, many of whom have specific use cases of their own in mind. Users are typically also contributors. People collaborate en masse, coming together to build a wonderful free geodata resource, and crucially it is open licensed and co-owned by everyone.
What is ‘authoritative/official’ data? Have you seen this data?
Maps created by mapping agencies. This is the traditional way of creating maps. A map making organisation, with map making professionals conducts the surveys and creates the maps. Often these are government, or government backed organisations. The data comes with a mark of authority because it is created by this organisation.
People often present “authoritative/official” geodata as the antithesis of VGI, but in fact it is produced in very similar ways, by humans who make judgements and also make mistakes, and at some stage there has been a decision to work towards a certain detail and accuracy level in representing the real world. There’s no such thing as a “completely accurate” map
Although not always the case, it’s worth noting that authoritative map data is very often not free. The standard old market driven approach is to license map data at great expense, and protect this business model through copyright enforcement. Exceptions to this include some open datasets from Ordnance Survey, and TIGER data in the U.S. In both cases “authoritative” data being released for free, but at a lower quality than other more expensive datasets. So “authoritative” does not necessarily mean non-free but also does not necessarily mean high quality.
Do you believe that there are more people using VGI maps in comparison to authoritative maps or do you believe that more people are using authoritative maps over VGI maps? Why do you believe this?
It is still the case that most maps that ordinary people encounter in everyday life are based on traditional authoritative data. VGI is very new, but large projects which release the data openly (I’m pretty much exclusively talking about OpenStreetMap here) are starting to have an impact and reach an increasingly mainstream user-base. We are seeing a shift towards end users seeing and using OpenStreetMap more and more.
If we think at the level of developers working with geo-data or just experimenting with geo-data in their bedrooms, there is a class of web developers and mobile app developers who make basic embedding use of raster maps. These are more numerous, and these are mostly still using Google Maps. But if we look at developers who are working with raw geo-data (not just basic embedding of raster), it’s quite likely we’ve already passed the point a long time ago where the majority of such developers are using OpenStreetMap data by virtue of its free and open availability.
If you had to choose between two different sources of data, would you choose a VGI dataset or an authoritative dataset? Why would you choose this option
I always try to use OpenStreetMap, because by using it you are supporting it. I use it when viewing maps on my phone, when printing maps, when emailing a link to a map, when embedding maps on websites I create.
OpenStreetMap exists to be used. That’s the goal. It creates a virtuous circle. Using it results in new people seeing and talking about it, and then some new people contributing to it. As an OpenStreetMap contributor, by using it myself I can help to spot areas of map data which can be improved. As an OpenStreetMap developer I can help to spot areas of the software tools and user experience which can be improved.
I choose to support OpenStreetMap because it is a wonderful open-licensed geodata resource which can benefit all of mankind. It is a not-for-profit good cause (This is not true of many other VGI projects I might contribute to)
Can you highlight any weaknesses of using VGI over authoritative data?
I think the weaknesses many people try to highlight are misplaced criticisms, or points which are outright incorrect. Let me give a few of these.
It is commonly held that VGI cannot be trusted compared to authoritative geodata. This issue of “trust” seems highly nebulous and subjective. I would argue that it actually boils down to a very indirect way of talking about data quality. No geodata is perfect, but if data quality is higher, more people will trust it, and OpenStreetMap data quality is ever-increasing.
People commonly criticise OpenStreetMap data shortcomings with a particular location in mind, but they should really fix it! (or at least point mappers to the location so we can tackle it, e.g. using osmbugs.org) With an open wiki-like process inviting anyone to improve the data, to criticise OpenStreetMap is to criticise yourself.
It is very common to hear people criticise aspects of the cartographic style presented on the ‘standard’ OpenStreetMap.org home page. This is a very visual thing which people are quick to notice, but it’s actually largely irrelevant. With open access to the underlying geodata (which OpenStreetMap offers for free) anyone can customise the cartographic style.
If I were to highlight a weakness, I would say one of the only fair criticisms of OpenStreetMap is the inability to achieve consistency across the dataset. OpenStreetMap currently has no upper limit on the level of detail volunteers can add, and this means that tremendous detail is added in one area, while another area is lacking. This weakness might mean that for some data use cases, a technically challenging process of smoothing over these imbalances can be necessary. For many use cases this is not a major shortcoming (and this is true of other issues of data quality)
What interests you about VGI?
I am excited by people coming together to collaborate on creating something great. I feel passionate about OpenStreetMap for this reason. The process and progress of map data being added is glorious and fascinating thing to behold.
VGI initiatives in general? I find many of them less worthwhile and subtractive from our global efforts. In particular many initiatives fail to open license and release the raw data which volunteers contribute. I would question the ethics of this exploitative practice. Hopefully potential contributors will see this and stay away, but this doesn’t always seem to work. I find it interesting that anyone would contribute to Google Map Maker.
What is required to produce high quality VGI within the UK?
There is no special requirement in the UK. OpenStreetMap’s approach to VGI was invented in the UK, but works worldwide, including creating the very first maps in the developing world for example. There are some differing considerations on a country by country basis. A key one is the availability of existing map data.
In the UK the Ordnance Survey still dominates provision of geo data. Many people in the UK are fiercely proud of our national mapping agency, but there is also a tremendous desire for open geodata. This gave rise to OpenStreetMap and continues to motivate volunteers to contribute in the UK. Partly in response to the “threat” of OpenStreetMap, the O.S. decided to release some of their lower quality datasets under a free open license. Nowadays we attract volunteers to OpenStreetMap showing that it can be better (mostly it is!) and more free compared to O.S.
What challenges do you feel would arise in the future of VGI?
I’ve already mentioned imbalances in level detail as a weakness we are struggling to tackle. This will become more of a challenge. Likewise other data issues such as vandalism and rogue importing will likely increase as the project grows, and we face challenges in structuring our project governance, but I think we will overcome these challenges within our community.
A big question is whether OpenStreetMap will remain relevant at all in the long run when faced with the challenge from other competing map providers. OpenStreetMap provides map data. It doesn’t attempt to compete on other features. There’s no OpenStreetMap aerial imagery and certainly no OpenStreetMap version of any 3D lidar photo synth features. It’s not something we are even *trying* to do, but If those things turn out to be the future, then OpenStreetMap might fade into irrelevance. This seems unlikely. Google streetview has been around for years, and hasn’t stopped people using normal maps for most use cases. Other forays into 3D have so far proved to have good gimmick value, but no long lasting effect on the way we use maps.
Another challenge might be if our form of vector map data can be auto-generated by some yet-to-be-invented machine learning OCR techniques. Of course competing crowd-sourcing initiatives might also be a challenge.
But there’s a certain glorious inevitability about the success of OpenStreetMap. It keeps getting bigger and better because people want open licensed map data. Even if OpenStreetMap somehow dies out, the data will live on with the same open license.
Do you feel that VGI is currently growing?
Yes. Massively so. In terms of quantity of data, and number of people taking part. See http://wiki.osm.org/Stats for some exponentially increasing curves.
How long do you feel VGI would be used for? Why do you believe this?
The data will be around, and will form the basis of interesting geo-experiments long into the future I’m sure. As a snapshot of the world as we see it today, OpenStreetMap is fascinating, because of the way it has be built by real people with local interests. But OpenStreetMap’s data is not being used to its full potential yet. The interesting question really is how long will it take for OpenStreetMap to really go mainstream?
How long do you feel authoritative data would be used for? Why do you believe this?
As mentioned, authoritative data currently forms basis of most maps in use today. I think this will continue until OpenStreetMap not only goes mainstream, but starts to push all other map data providers out of the market. I’m not sure if this will ever happen (I think in ten years time we’ll know either way) but in any case OpenStreetMap is adding value, and can add much more value, alongside authoritative map data.
Where do you see VGI in five years from now?
Impossible to say. We’re at an exciting juncture right now. In five years OpenStreetMap could be massive, or it could be coasting along still yapping at the heals of other map providers.
What do you believe the future trends are for VGI?
We’ll see more commercial propositions built on top of OpenStreetMap, and I think this will help to drive things forwards. We may see the emergence of a new kind of “authoritative” data, built on top of VGI. Map data authorities could take a snapshot version and “bless” it as trustworthy, or perform some elaborate branching of the dataset to arrive at an authoritative version.
Is there anything else you would like to add about VGI and its future trends?
VGI / crowd-sourcing initiatives should open-license the data they gather, to provide it back to those who contribute. In fact it should be regarded as unethical not to do so, and we must campaign strongly against instances of closed data crowd-sourcing (such as Google Map Maker) to ensure that this exploitative practice does not become a trend.
Open licensing is about giving the data back to your contributors (which should help you attract them in the first place) but it’s also about data sharing *between* different initiatives, and ensure your data gets used as widely as possible. New VGI initiatives should also consider the compatibility of their open license with that of OpenStreetMap. How might we share data? Or could OpenStreetMap be a good platform for directly publishing the data? By doing this you can be taking part in the largest VGI initiative of them all!