>A simple, beautiful, semwebby, linked data-enabled
I can certainly picture what "linked data-enabled" would mean for each of those apps; what would "semwebby" mean?

Obviously there are completely functional applications in each of these categories that exist outside of the Sem Web arena. Would really like to understand the benefit that Sem Web/Linked Data would bring to each of these bullets.

Bob, with semwebby, I mean RDF-based, with easy data import/export and basic resource consolidation. I added the "LD-enabled" later which I guess is mostly redundant but highlights the idea of maybe having a separate page or web-accessible graph for each entity used by the apps.

Eric, the benefit isn't necessarily having SemWeb technology at each individual app, but it's an enabler for being able to easily integrate and work on top of the combined data later.

As discussed on IRC, many of these features already exist, and the rest could be implemented without too much difficulty, with ODS (OpenLink Data Spaces). I now understand that what you meant by this post was that you want to build such a suite of applications, not find one ... but perhaps others won't want to wait for your creation, and equally possible, perhaps you'll want something to compare your own efforts to and (possibly) improve upon...

I'm not sure what you mean by RDF-based. spam snip The look-and-feel of the ODS application suite may not be to your liking, but that's more of a skinning question than a functionality question -- the RDF, SPARQL, and other features you've listed here are all baked in.

"depth first" is definitely something the semantic web is lacking at the moment. once google maps was there, creating kml wasn't much of a problem anymore. there was just enough pull.

Ted, maybe your comments show the different philosophies quite nicely. OpenLink focuses on features and considers the UI as "the rest" and "a skinning question" where I think the UI is the most important thing and possibly the RDF community's main challenge these days. Maybe I read too much 37signals stuff, but I agree with them that we have to stop thinking that an app's usability can be added after it was programmed.

Robert, couldn't agree more. I hope that things will improve now that a stable set of essential specs is in place and a lot of infrastructure problems have been solved sufficiently.

Benjamin - we're working on a "simple, beautiful, semwebby, linked data-enabled" wiki at www.swirrl.com. It's quite a challenge to combine all these things :-) We've tried to concentrate on keeping it simple and beautiful, and add the semwebbiness bit by bit. In fact we've recently taken out some of our first attempt at the sem web stuff because it wasn't simple enough.

We'd like to use the semantics to improve things for the user without them having to be a data modeller. So there's a lot of stuff still on our to-do list, but we store structured data as RDF, alongside regular wiki pages and expose that linked-data style. ie every item has an HTML version for the humans and an RDF version for the machines and semantic web enthusiasts :-) with all the appropriate content negotiation. But we've got a way to go in terms of ontology features, owl:sameAs, linking out to existing LOD etc. If you get a chance to try it out, I'd be pleased to hear what you think of it.

+1 to all of this.

I think Freebase shows the two sides to this way of thinking well. They have millions of topics and some generic data tools. They've had to add in specialized data tools, like typewriter, to perform specific tasks in a user friendly manner.

Feed reader - hey, if it eats RSS 1.0, does that count?
I can see this:
 * Speaking to an archive service (ie, the Semantic Web Google Desktop) and providing 'documents' with dc:titles for it
* Simple SPARQL services - show me articles published on date X by author Y

Bug tracker / TODO:
launchpad has always been neat-o-rama; I wish it were baked into trac (sparql, rdf output).

Address book - uhm, didn't you make knowee already :)

online profile generated from all my data - knowee, mybloglog (outputs foaf)?


Stuff I want:
 * I want my instant messenger to talk to my address book and reconcile who's who.
 * I want my instant messenger to be able to talk to other people's computers, via Jabber, so I can SPARQL them if I want to.
 * I want the Kevin Bacon agent, which just like on facebook suggests 'people you might know', by recursively indexing the address books of my friends and smushing.

 * I want a desktop triplestore of some description, which, with one click, I can *publish to a public site* with public data + private data (https + basic auth ftw).
   * I don't care if its actually hosted somewhere else, and my local computer just connects to it
   * I want this thing to be the backbone of every application on my computer that wants to store some data.

More:

If I'm reading something, anywhere, my on-screen agent should start showing me related things. Especially my related private data. E.g. if someone IMs me and mentions another person, show the contact info for that other person (and recent messages between her and me). Who else knows both the IM partner and the mentioned person? Got any photos with all of us?

If I'm composing an IM or email and I start to include a date, immediately show me my calendar for that date. That way I can notice a conflict and change my suggestion before I even send it. But when I do send the message with some date, that's now associated with my calendar. (And don't just *copy it* to the calendar, gmail!)

If I type a method name in my code, show me the complete docs for that method, and its source code, and every other time I've ever made a call to that method.

I think I have a common pattern of complaining about an apparent bug to someone over IM. In some cases, I later decide to file a ticket about the bug. That's when I want to be reminded (passively, like all of these proposals) of any IM conversations. Maybe this is just a special case of *always* digging up similar IM conversations whenever I'm composing anything.

I work on home automation projects, and these should be tied in too. When I'm writing to someone, show me the last few times my cell phone detected his cell phone's bluetooth id. Add a rule (in n3): If it's below 50 deg, it's after 7pm, my calendar doesn't show me going out, and my work computer has been idle for more than 5 min, assume I'm on my way home and turn on the house heater (for 15min max, then give up).

related: why email clients need to change