When I first started with computers, I took naturally to programming. Over the years, I've rarely spent more than a month or so without programming anything. In 2013, I got a chance thanks to support from the Shuttleworth Foundation to realise my ideas for provenance of digital works, which I did, with the help of others, in elog.io (since then shut down, but source code available). Not having touched elog.io for about half a year (but still keep getting prizes and nominations for that work!), I've had a chance to think a little bit more about that work and what I would've done differently. To sum up the most important learnings:
- I would've put a focus on making elog.io a distributed database to start with. It sort of is, under the hood, and the intention is there, but it would've been amazing to actually build a distributed database for provenance information. This is something that's dearly needed in the world.
- I would've focused on the API, rather than the UI. Getting the mechanics to work before doing anything else.
- I would've focused more on creating tools to help people work with the information.
- I would've split data sources between provenance information and relations.
- I would've focused the provenance information on telling the story of a work, not (only) its license.
- I would've enabled user contributions of information already from the start.
This would take some additional work and thinking, but overall, I'm thinking that provenance information can, if standardised, be better conveyed through and with the semantic web: we have the standards already in place, and what's needed is just some tools to make it easier to work with. This would be distributed and easily done, even if it would take longer to scale. What became elog.io, I would've shifted to focus only on the relations between objects: between different works, and between the same works, ie. identifying an image on Flickr as the same image as one on Wikimedia Commons. That's the sort of thing that I can see a clear benefit in curating separately from the works themselves, and it would allow people to contribute to the knowledge about creative works by creating associations.
All of this would take a wee bit more of thought regarding how to actually implement it and how to make it useful, but I do believe that this also reflects in a way some of the failure we had while doing elog.io: not putting the above principles first simply meant that it was difficult for others to engage in the work and contribute, and it ultimately made elog.io less useful than it could've been.