Georgi Kobilarov has a refreshing post, suggesting Making Linked Data work isnt the problem. I'm inclined to agree with most of what he says. The technology in itself isn't a solution to any problem, rather an enabler to solve problems. While the idea of serendipity is appealing, it isn't very good justification for a huge global commitment of resources. So what kind of problems do we, as living, social and technological organisms wish to solve?
To start exploring this space I reckon there are (at least) two general modes of knowledge use. The first is relatively domain-specific, directed by a set of requirements associated with a corresponding set of real-world tasks and operations. These I'd put under the umbrella of Applications, akin to the computer applications we already use but augmented with knowledge engineering facilities and access to the Web of Data. As a shortcut the starting point here is Connolly's Bane: "The bane of my existence is doing things I know the computer could do for me.". But in general it goes far further, in that there are plenty of beneficial things we don't already do. A second mode would be ad hoc, fairly immediate, unplanned, call it Just-in-time problem solving, the kind of thing that we currently turn to search engines for.
As an example of the Applications mode, one of the early drivers for the Web was e-commerce. I think I'm fairly safe in saying that only the surface of the potential there has been scratched. There's a hint of what can be possible with things like the individual-targetting of Google Ads and Amazon recommendations. In this space the GoodRelations ontology is a marvellous baseline. But what we're not really seeing yet is the whole supply chain from the manufacturer to consumer being integrated. Fairly loosely-coupled (as it is today) In one direction there are the financial aspects ("follow the money"), in the other direction is all the transport, manufacturing and processing that go from raw materials to delivered finished product. Within those different parts of the pipeline there are a whole host of problems relating to technology tied together by human and natural resources.
Alongside this commercial world there are macroeconomic and macrosocial systems, those areas traditionally covered by government. We're already seeing some movement around transparency with the various government data projects, but I think we're still a very long way from seeing genuinely informed policy and decision making. Reflecting the darker side of advertising right down to commercial spam and taking advantage of general ignorance, good governance is seriously compromised by self-interest (of individuals and corporations) and misinformation. I recently heard a radio programme talking about the UK Conservative Party's successful "Broken Britain" election campaign. An aspect of this was that violent crime was perceived as being on the increase. However the actual statistics suggest that in reality this malaise had actually been declining (see Murder rate lowest for 12 years "Home Office figures show overall crime fell by 5% in England and Wales"). Politicians will always lie, but damage is only done when they get away with it and aren't held to account with the facts. But I don't really want to suggest that prevention of political badness is the goal here, rather the encouragement and facilitation of goodness (man...).
Another huge area where there are countless problems to solve is science. While the Web has vastly improved information sharing and been a boon to research, I'm not sure the underlying methodologies have changed that much. I'm convinced the open sharing of knowledge at the data level can offer A New Kind of Science (no hyperbole there!).
There are plenty of other application domains that could benefit from a bit of Web-scale knowledge engineering. Ok, I'll name one more bundle: the Arts.
Ok, moving on to the Just-in-time mode of problem solving, take a look at the following list (random stuff that came off the top of my head when I woke up this morning). Imagine how you would solve these problems now, and then think how you might solve them with a thousand programmers at your beck and call. Most of them need something considerably deeper than a keyword/linkrank document search. I've dumped this list over on the ESW Wiki, additions and discussion welcome over there (I still haven't implemented comments on this blog, so if you have a comment for anything either mail me or blog it (and mail me) or tweet or use Facebook...).
- I'd like to upgrade the computer I use for video editing. My budget is about 300 euro. What should I buy?
- Who should I get to make the soundtrack to my new film?
- I've bought an Ubuntu laptop to replace my old Apple, I'd like it to run applications that fulfil all the tasks I have on the old machine. What do I need?
- Should HTML use namespace prefixes?
- Is there a political motivation behind Royal Weddings?
- Who should I vote for?
- Who might make a good (romantic) partner?
- I wish to sell my double glazing products in sub-Saharan Africa, who should I contact?
- Who might make a good (business) partner there?
- I got a mail from someone claiming to be my cousin, asking for a loan. Should I give them the loan?
- I've got an interesting rash. Should I see a doctor?
- I wish to enlarge my penis. What method is safe and reliable?
(Sorry, couldn't resist the last one - but it's a valid example of where you'd need good healthcare data alongside reputation and provenance information)
PS. danbri points me to a short 1989/90 document which contains a fairly similar list (minus references to genitalia) : Information Management: A Proposal, by a certain Tim Berners-Lee. Go read it. Now!