Every year I enjoy writing a post with my predictions for the next year.  It's a nice way to empty the old thoughts from my brains so there's room for new ones.  Plus I leave a trail of blog posts so I can see how my thoughts change from year to year.

In retrospect I've realized that each year tends to have a theme or two, and that the hardest part about making accurate predictions for a given year is identifying the themes that will drive it.

In 2007 the driving themes were social networking and online entertainment.  In 2008 the themes were distributed and mobile communication, with a dash of cloud computing sprinkled on top.

The theme for 2009 is almost absurdly easy to identify:  the economy.  If you thought 2008 was about the economy, just wait for 2009.  You truly ain't seen nothin' yet.  A deflationary black hole is sucking all of the money out of the economy and we haven't even seen the impact yet.

Every time I hear somebody talk about how the market has bottomed and the economy is starting to improve I mentally picture them in this position:

So if we're talking about the economy and you see me chuckle, you know why.

Hope is a great mindset to have and a fantastic slogan for winning Presidential campaigns apparently, but it is not such a great lens to view reality thru when you're trying to make money.

A post by Erick Schonfeld at TechCrunch today really got me riled up:  The SEC has shut down Prosper, a peer-to-peer lending site.  This was up in the air until yesterday:

Yesterday, the SEC issued its formal cease-and-desist letter (embedded below or download PDF), outlining its reasoning for characterizing Prosper as a seller of investment, something prosper had vigorously resisted in the past by arguing that it was merely a marketplace matching lenders and borrowers. But the SEC is having none of that.

If this sounds familiar, it's because this is an exact rerun of what happened with the original Napster and the music industry, only worse in my opinion.

The key here is that Prosper itself was not lending or borrowing, it was simply matching up willing borrowers and willing borrowers.  It also provided additional services such as collection and tracking.  The HORROR.

The real fact is, if private citizens were allowed to freely lend to one another, the private banking cartel that is our central banking system would lose the little control they have over the economy.  The free  market would freely set interest rates and people and businesses would be free to do an end-run around our corrupt and bloated financial system.  The financial engineering that has allowed Wall Street to siphon off trillions of dollars in profit at our expense would be crippled.  The SEC is simply acting as the enforcement arm of our private national banking cartel.

Don't fall under the protection of the cartel?  Goodnight, chump.

...is "How can this technology make me money?"

Notice that I say this is the question they must learn to answer, not what I personally am asking.  That's because there is no answer that will work for everyone asking the question, and I already know the answer for me.  People who want to evangelize this technology must learn to make the asker realize why this is not the correct question to ask.  As I said before, it's a mindset problem.

You also need an elevator pitch for this answer.  Something you can say in under 30 seconds.  If the answer takes more time than that you won't be able to sell it.

One of the questions I'm being asked somewhat frequently these days, probably because I write about it a lot, is "what can the Semantic Web do for this product/idea/whatever?"

My reaction is always a brief pause while I realize that the person asking the question doesn't really understand what it's about.  In the end, they're the one who has to figure that out.  The short answer is "it depends on what you're building".  It's like asking "What will learning how to read and write do for me?"  It depends on what you want to do with it, trying to ask somebody else is pointless.

I think this is ultimately a mindset problem.  Relational databases have placed limitations on software development that have been in place for so long that people aren't even conscious of them, they're just accepted.  Like the earth being flat.  Or the "fact" that  man cannot fly.  Software engineers have these limitations drilled into their heads from the time they're in high school until they graduate college, and then they perpetuate these ideas.  When a couple of guys sit around drinking a beer and drawing flowcharts on bar napkins, they have the limitations of a relational database in mind without even consciously thinking about it.

People think of the Semantic Web/Linked Data Web as something tangible and easy to point at, like rounded corners, or AJAX, or software that runs in a browser.  Or, more often than not, something "better than Google" in a nebulous kind of way.  It's not.

In my brain, I think of it as a direct replacement for relational databases.  This is from a programmer's perspective, of course.  I view it as something that you will plug in in place of the relational database that used to hold all of your stuff.  It is a means to let your applications use the Internet itself as its database.

I don't know what connections between your data and the outside world make sense, unless I'm involved in brainstorming the product.  But if you use a relational database you can't make any of those connections, they're precluded by your platform.  There's a ton of useful data out there for you to connect to, but it's up to you to figure out what you need.  You must know where you're going in order to get there.

To me, the Semantic Web is a fundamental shift in software architecture.  I've said this before, and I'm still convinced of it:  I will never build another application on a relational database, even if I don't plan to use any outside data.  Why would I consciously cripple my application?

UPDATE:  Kingsley has reminded me that there is indeed a line that needs to be drawn between the RDF store and the old relational database.  There is, already, middleware available that will facilitate mixing and matching RDF and relational databases in cases where the relational database is already entrenched.  I have corrected the flowchart :)

I've been playing with Twitter for the past few days on Trent Adams' recommendation.  Here's my profile.  I never really saw the difference between Twitter and a blog plus RSS, but he said there are some good conversations that happen there which you'll miss if you're not a part of it.  And, after a few days of play time I'm actually starting to enjoy it.

What it ends up being is a personal online chat room where you pick the participants.  You need a fat client, similar to a modified instant messaging client, or it's basically useless because you don't get the realtime updates, which kills the deal.  I use Twhirl.  There's a lot of noise (for example, I tweeted (twitted?) about how much the hotel coffee sucks this morning), but it's actually pretty cool when it works as an online conversation.  I see a lot of potential in this communication model for enterprise teams.

However, Twitter was not built as a chat platform using XMPP or any of the other actual chat protocols that are available.  It was built as a micro-blogging platform using a RESTful interface, and it does not lend itself well to scaling to a vast number of users bombarding it with requests for updates.  It goes down, a lot.  Just goes to show you that buzzword technology still needs to be applied judiciously.

Twitter is a perfect candidate to switch to an RDF back-end for publishing twits.  They could make this change in a week and while the clients wouldn't automatically switch over to SPARQL querying they'd at least have a scalable back-end going forward.

Oh, and I have no idea how they could possibly monetize this thing.  This type of service can not live forever as a private service, it can only work long-term in a distributed, decentralized model.  I have a feeling it's just a glimpse into the future of messaging.

Imagine you're sitting in an interview and your phone suddenly tells you that your friend John went to college with the person you are interviewing with.  Pretty cool huh?

I was just complaining the other day that radical innovation seems to be in short supply these days.  Then today I stumbled across this really cool idea called Mobile FOAF (they're calling it FoafMobile, but that sounds too much like a funny little car to me--maybe because it rhymes with PopeMobile?--so I'm calling it Mobile FOAF instead).  The basic idea is that friends and people you know, who are in close physical proximity to you, can be discovered using Bluetooth-compatible mobile devices.  It's a small world, after all...

The gist of it is that each Bluetooth device has a unique identifying address (like a MAC address), and if you put that address in somebody's FOAF graph you can trace a Bluetooth device back to its owner.  You could query your FOAF graph for anyone you know who has that particular Bluetooth device, and even ask your phone to show you if anyone in the restaurant knows any of your friends.  If so you could pull up their name and picture and go find them to have a cup of coffee.  A magician could have a field day with this stuff.

The writeup I found on Mobile FOAF is actually pretty old (2003), but I'm not sure how something like this would have been useful before SPARQL was ratified this year anyway.  <hint>Hopefully this will get picked up by some enterprising company and taken to market,</hint> because this would certainly be some radical innovation.  If a Bluetooth "beacon" could be baked into wireless routers you wouldn't even need GPS to locate somebody...

The Web is still in its infancy.  It sounds like an absurd claim, except that the person making it is Sir Tim Berners-Lee, the guy who invented the Web in the first place.  I happen to agree with him.

The first incarnation of the Web changed the world, and what we call "Web 2.0" was only an incremental improvement on that.  It made the baby-Web usable for normal people.

I've had the pleasure of speaking with Tim on a few occasions and he is patiently waiting for everyone to get over the love-fest with what they currently know as the Web so we can focus on moving things forward and innovating again.

The end-game here, and what Tim has had in mind from the beginning, amounts to a completely new paradigm in software.  It's about creating an Internet-wide mesh of data which a given software application can use as easily as its own database. 

Fortunately, the Web is hurtling towards adolescence.  SPARQL is the glue that makes this global data mesh usable, and it was just ratified this year.  The Web's voice just cracked.

While advances like service-oriented architecture were steps in the right direction, they were still just an incremental improvement over client-server based architectures.  And as useful as that is, I think an unfortunate side-effect is that an entire generation of programmers was raised with their thinking firmly locked in the old paradigm.  I've tried to evangelize the new paradigm myself, and I can attest that it's like trying to convince people that man can actually fly using airplanes while they turn around and go back to work on pimping out their horse and buggy.

Tim's vision of the Web is the ultimate logical evolution of Web technology.  Fortunately there are some recent developments that indicate that the Web has hit a growth spurt and we may soon see another explosion of innovation based on the Next-Generation Web.

One of the most unfortunate things to happen to the Semantic Web movement is its name.  The word "semantic" always, ALWAYS, gets people thinking of semantic text analysis--that is, analyzing a chunk of text and figuring out what it's about.  That's what Google is for, that is NOT what the Semantic Web is for.

Tim Berners-Lee does a good job of pointing this out in a recent post where he discusses the killer application for the Semantic Web:

Text search engines are of course good for searching the text in documents, but the Semantic Web isn't text documents, it is data. It isn't obvious what the killer apps will be - there are many contenders

What's really annoying is that even when I go to Semantic Web meetups at MIT, people there for some reason are stuck on building a better Google.  You hear it all the time, and it's just really off base.  Tim feels the same way:

One thing to always remember is that the Web of the future will have BOTH documents and data. The Semantic Web will not supersede the current Web. They will coexist. The techniques for searching and surfing the different aspects will be different but will connect. Text search engines don't have to go out of fashion.

I've been saying this for a while:  Why in the world would you try to build a better Google?  It already does an OUTSTANDING job of doing what it does--searching text.  No thanks, I will keep my Google, what we REALLY need to address is identifying objects properly on the Web--linking data.

Kingsley and some other people have been pushing the Linked Open Data brand instead of Semantic Web lately, and I will agree that it fits much better.  The word "semantic" just throws many people way off course, and I think it actually creates disinterest in a lot of people who would otherwise dive in and start using it.  "Linked Open Data" MUCH more accurately describes what it's about.

Linking data is about exposing data so that it can be re-used in other ways that you hadn't thought of yet and queried using SPARQL.  Its analogous to Service Oriented Architecture in that it exposes part of your application to be re-used and mashed-up across the entire Internet--except that it's exposing the data, not the functionality.  (But wow, put the two together and you REALLY have something cool!)

By the way, here's the money shot from Tim if you're looking to make money from the Linked Open Data/Semantic Web, print this one out and pin it to your wall:

So if you are a VC funder or a journalist and some project is being sold to you as a Semantic Web project, ask how it gets extra re-use of data, by people who would not normally have access to it, or in ways for which it was not originally designed. Does it use standards? Is it available in RDF? Is there a SPARQL server?

In Malcom Gladwell’s book The Tipping Point, he describes a certain class of people as “connectors”.  They’re people who know everyone, and thus are capable of spreading information and ideas quickly.  Social networks hint at leveraging these personal connections in digital form, but so far have failed to do much with it.

LinkedIn is the only application I’ve seen that really explores this idea from a social networking perspective.  Your network, the number of people you have access to, grows exponentially with each relationship you add.  And if you connect with one of those "connectors" your network will grow even faster (because you become connected to the people they know).

While it makes for an interesting interface and a fun way to explore social connections, LinkedIn--and all the other social networks I've seen--really don't do much with this connection data.  You can ask your network a question or see who’s hiring in your network, but it’s mostly just for ego gratification.

OpenID and FOAF change all that.  All of a sudden people become ACTUAL connectors, connecting together pools of meaningful, rich data that applications can use. 

Two applications which speak OpenID and FOAF all of a sudden have a common data element, a JOIN in SQL-speak.  It's possible because OpenID turns a person into a URI, which naturally lends itself to be used in FOAF.  FOAF is just a list of people, after all.

This is really cool stuff.  It’s automatic data integration across the Internet, using people as connectors. 

This is reason why Kingsley and a bunch of other people like to call the "Semantic Web" the "Linked Data Web".  Potayto-potato, it's all the same to me.  It’s cool, though.  It lets an application traverse the social graph to do its thing instead of being confined to its own network.  It allows an application on one network to access Person C’s data, on another network, by going from Person A to Person B to Person C, and then to their data.

I thought (still think) this was really cool when I finally understood it, and I don’t think many people have an appreciation for the opportunities this opens up.  Although, with the recent traction OpenID has been getting I think it's only a matter of time.

P.S.  This data integration using humans as connectors won't work via proprietary API's or microformats, by the way, because the enabling technology for this is SPARQL, which requires RDF to work its mojo.  Exposing proprietary API's means that you have to custom duct-tape together the data using middleware.

One of the interesting debates from the Semantic Web meetup this month, which has stuck in my head ever since, is the argument over whether it's worth it to try to compile a master, global, schema--ontology--vocabulary--whatever you want to call it.

I could tell that the academic side of the crowd was again' it, but the business side of the crowd was 'fer it.

Guess which side will win?

The analogy that was given (sorry don't remember who said it) was that a pig farmer might use a different concept of "a time" than I might.  So 3PM Pig-Farming Time might equal 2PM Jason Time.  (If I got the analogy wrong I'm sure I'll be corrected, but it was something like that.)  Or, one that kept coming up as a real-world example was that a transaction has totally different meanings across industries.

I think that thinking along those lines is misguided.  You can't define everything in the world, but you can sure define a lot of it.  That's what businesses end up spending a lot of their time doing--defining their world.  I am intimately familiar with this, as it is one of the biggest problems that we struggled with at Latigent (now Cisco).  Company A will define "revenue" as one thing, Company B will define it as something else. 

This isn't an either/or proposition.

You can give them both definitions, they really don't care..  This is an instance where you really can have your cake and eat it too.  As long as both values are populated, they will take the one they need.

To me, this is one of the huge value-adds of the Semantic Web.  You can go out and see if somebody has already developed a schema for what you're trying to build, and if they have you don't have to write it yourself.  And your app will automatically inter-operate with theirs, in some areas.  It's really cool.

Every company out there should really be using available schemas as much as possible--get into the Semantic Web game early and often.  All you have to do is start adopting a standard schema, taking as much as possible from what's already publicly available.

Why NOT build a giant global schema?  Isn't that what the ontology-building stuff is all about anyway?  It's not like it all has to be from the same place.  If I come up with the best definition of a "location", because mine includes height above sea level and active quantum dimension, you're free to use that in your application for "location" and develop some new schema pieces for the custom stuff you're building.  Eventually the global schema will just kind of materialize out of the best (well, technically probably the most POPULAR) pieces of publicly-available schema out there anyway.

Heh, boy I can't even imagine what this is going to do to the business intelligence space eventually.  Totally demolish and rebuild it, is my guess.