What is it that makes my favorite online services so compelling? I’m talking about the whole family of services that includes Dopplr, Wesabe, Twitter, Flickr, and del.icio.us among others.

I find it interesting that people don’t generally refer to any of these as “web sites”. They are “services”.

I was fortunate enough to spend some time with Dopplr’s Matt Biddulph and Matt Jones last week while in London where they described the architecture of what they’ve built in terms of connected data keys. The job of Dopplr, Mr. Jones said, was to “surface coincidence”.

I think that term slipped out accidentally, but I love it. What does it mean to “surface coincidence”?

It starts by enabling people to manufacture the circumstances by which coincidence becomes at least meaningful if not actually useful. Or, as Jon Udell put it years ago now when comparing Internet data signals to cellular biology:

“It looks like serendipity, and in a way it is, but it’s manufactured serendipity.”

All these services allow me to manage fragments of my life without requiring burdensome tasks. They all let me take my data wherever I want. They all enhance my data by connecting it to more data. They all make my data relevant in the context of a larger community.

When my life fragments are managed by an intelligent service, then that service can make observations about my data on my behalf.

Dopplr can show me when a distant friend will be near and vice versa. Twitter can show me what my friends are doing right now. Wesabe can show me what others have learned about saving money at the places where I spend my money. Among many other things Flickr can show me how to look differently at the things I see when I take photos. And del.icio.us can show me things that my friends are reading every day.

There are many many behaviors both implicit and explicit that could be managed using this formula or what is starting to look like a successful formula, anyhow. Someone could capture, manage and enhance the things that I find funny, the things I hate, the things at home I’m trying to get rid of, the things I accomplished at work today, the political issues I support, etc.

But just collecting, managing and enhancing my life fragments isn’t enough. And I think what Matt Jones said is a really important part of how you make data come to life.

You can make information accessible and even fun. You can make the vast pool feel manageable and usable. You can make people feel connected.

And when you can create meaning in people’s lives, you create deep loyalty. That loyalty can be the foundation of larger businesses powered by advertising or subscriptions or affiliate networks or whatever.

The result of surfacing coincidence is a meaningful action. And those actions are where business value is created.

Wikipedia defines coincidence as follows:

“Coincidence is the noteworthy alignment of two or more events or circumstances without obvious causal connection.”

This is, of course, similar and related to the definition of serendipity:

“Serendipity is the effect by which one accidentally discovers something fortunate, especially while looking for something else entirely.”

You might say that this is a criteria against which any new online service should be measured. Though it’s probably so core to getting things right that every other consideration in building a new online service needs to support it.

It’s probably THE criteria.

One of the more interesting sciences, in my mind, is how information relevance is both determined, surfaced and then evolved.

In Fred Wilson’s recent Cautionary Techmeme Tale he argues that making news popular takes away its social context and therefore becomes meaningless. He found Techmeme more useful when its sources more closely resembled his network of friends:

“For years, I’ve been using curators to filter my web experience…Techmeme has been the killer social media curator for my world of tech blogs. Lore has it that it was created using Scoble’s OPML file. It doesn’t matter to me if that’s true or not, I love that story. Because my OPML file was unusable until I found Techmeme and after that I stopped reading feeds and started reading curated feeds.”

This feeds into a larger argument about why pop culture and the art of being or becoming popular can be a bad thing. Not long ago I was inspired by the movie “Good Night and Good Luck” to dive into this idea myself:

“The real problem with popularity-driven models is that they reduce both the breadth and depth of the sources, topics and viewpoints being expressed across a community. Popularity-driven models water down the value in those hard-to-find nuggets. They normalize coverage and create new power structures that interesting things have to fight through.”

This is exactly why personalization, recommendations and social media technologies really matter. They can solve this problem of creating conformist media consumption practices by creating relevance through networks of people rather than through networks of commercial institutions.

I haven’t used My Yahoo! as much as I’d like, but there is a simple function in it that I love which could ultimately create amazing benefits for people who want a human filter for the Internet. It’s called “Top Picks”.

“The Top Picks module automatically highlights stories from your page, based on the articles you have recently read on My Yahoo! The more stories you click on, the more you will see this module reflect your interests.”

Actually, the technology beneath it is not so ’simple’ but the application of it here makes so much sense that it feels like it’s simple when you watch it work. It works by using implicit behaviors. I don’t have to tell it what I like. It learns.

If it could also show me what my social network is tapped into right now, then the experience would feel nearly complete.

Media researchers will note here that people need pop culture to feel connected to a greater whole. I believe that’s true, too. Television is an amazingly powerful community builder.

But I would gladly trade a powerful singular social voice tied together by networks of distribution ownership for a less unified but still loosely connected network of pop culture tied together by my personal activities and my social connections.

Jay Rosen has an interesting post on the failure of AssignmentZero, an effort to build a publicly funded crowdsourced news organization.

Among the many lessons, he keeps coming back to motivation and incentive.

“A well managed project correctly estimates what motivates people to join in, what the various rewards are for participants, and where the practical limits of their involvement lie.

…amateur production will never replace the system of paid correspondents. It only springs to life when people are motivated enough to self-assign and follow through.”

The idea wasn’t fundamentally broken, in my mind. Crowdsourced news is very powerful. As Derek Powazek said,

“At its best, crowdsourcing is about expanding the walls of the newsroom to the internet, giving an opportunity to people with real experience to share their expertise. This is a point that’s often lost on people who are just looking to make a quick buck on Web 2.0.”

More than anything else, I suspect that AssignmentZero failed because there weren’t any readers. Motivation wouldn’t have been a problem with a NYTimes-sized audience.

To date, I’ve never seen a better explanation of the motivations in collaborative online experiences than Yochai Benkler’s paper called Coase’s Penguin. One of my favorite excerpts from that is where he warns against paying for contributions from the community:

“An act of love drastically changes meaning when one person offers the other money at its end, and a dinner party guest who will take out a checkbook at the end of dinner instead of bringing flowers or a bottle of wine at the beginning will likely never be invited again.”

There are as many motivations as there are contributors in a shared media project. What holds them together is more art than science. Some of that art includes good timing and luck. But it also requires a unique kind of commitment and salesmanship from the leaders of the project.

I’ve begun to wonder if the tipping point happens when the confluence of the community size, the ROI to the contributors and the depth of the trust relationship with the company or the brand creates more value than the sum of the parts. Maybe the science of collaboration services can be found by quantifying the meaning of the relationships between those elements: size, cost, benefit and trust.

Or it could also be that the secret sauce inside the Craig Newmarks, Stewart Butterfields and Jimmy Waleses of the world is much more complicated and nuanced than anyone realizes.

The Internet platform business has some unique challenges. It’s very tempting to adopt known models to make sense of it, like the PC business, for example, and think of the Internet platform like an operating system.

The similarities are hard to deny, and who wouldn’t want to control the operating system of the Internet?

In 2005, Jason Kottke proposed a vision for the “WebOS” where users could control their experience with tools that leveraged a combination of local storage and a local server, networked services and rich clients.

“Applications developed for this hypothetical platform have some powerful advantages. Because they run in a Web browser, these applications are cross platform, just like Web apps such as Gmail, Basecamp, and Salesforce.com. You don’t need to be on a specific machine with a specific OS…you just need a browser + local Web server to access your favorite data and apps.”

Prior to that post, Nick Carr offered a view on the role of the browser that surely resonated with the OS perspective for the Internet:

“Forget the traditional user interface. The looming battle in the information technology business is over control of the utility interface…Control over the utility interface will provide an IT vendor with the kind of power that Microsoft has long held through its control of the PC user interface.”

He also responded later to Kottke’s vision saying that the reliance on local web and storage services on a user’s PC may be unnecessary:

“Your personal desktop, residing entirely on a distant server, will be easily accessible from any device wherever you go. Personal computing will have broken free of the personal computer.”

But the client layer is merely a piece of the much larger puzzle, in my opinon.

Dare Obasanjo more recently broke down the different ideas of what “Cloud OS” might mean:

“I think it is a good idea for people to have a clear idea of what they are talking about when they throw around terms like “cloud OS” or “cloud platform” so we don’t end up with another useless term like SOA which means a different thing to each person who talks about it. Below are the three main ideas people often identify as a “Web OS”, “cloud OS” or “cloud platform” and examples of companies executing on that vision.”

He defines them as follows:

1. WIMP Desktop Environment Implemented as a Rich Internet Application (The YouOS Strategy)
2. Platform for Building Web-based Applications (The Amazon Strategy)
3. Web-based Applications and APIs for Integrating with Them (The Google Strategy)

The OS metaphor has lots of powerful implications for business models, as we’ve seen on the PC. The operating system in a PC controls all the connections from the application user experience through the filesystem down through the computer hardware itself out to the interaction with peripheral services. Being the omniscient hub makes the operating system a very effective taxman for every service in the stack. And from there, the revenue streams become very easy to enable and enforce.

But the OS metaphor implies a command-and-control dynamic that doesn’t really work in a global network controlled only by protocols.

Internet software and media businesses don’t have an equivilent choke point. There’s no single processor or function or service that controls the Internet experience. There’s no one technology or one company that owns distribution.

There are lots of stacks that do have choke points on the Internet. And there are choke points that have tremendous value and leverage. Some are built purely and intentionally on top of a distribution point such as the iPod on iTunes, for example.

But no single distribution center touches all the points in any stack. The Internet business is fundamentally made of data vectors, not operational stacks.

Jeremy Zawodny shed light on this concept for me using building construction analogies.

He noted that my building contractor doesn’t exclusively buy Makita or DeWalt or Ryobi tools, though some tools make more sense in bundles. He buys the tool that is best for the job and what he needs.

My contractor doesn’t employ plumbers, roofers and electricians himself. Rather he maintains a network of favorite providers who will serve different needs on different jobs.

He provides value to me as an experienced distribution and aggregation point, but I am not exclusively tied to using him for everything I want to do with my house, either.

Similarly, the Internet market is a network of services. The trick to understanding what the business model looks like is figuring out how to open and connect services in ways that add value to the business.

In a precient viewpoint from 2002 about the Internet platform business, Tim O’Reilly explained why a company that has a large and valuable data store should open it up to the wider network:

“If they don’t ride the horse in the direction it’s going, it will run away from them. The companies that “grasp the nettle firmly” (as my English mother likes to say) will reap the benefits of greater control over their future than those who simply wait for events to overtake them.

There are a number of ways for a company to get benefits out of providing data to remote programmers:

Revenue. The brute force approach imposes costs both on the company whose data is being spidered and on the company doing the spidering. A simple API that makes the operation faster and more efficient is worth money. What’s more, it opens up whole new markets. Amazon-powered library catalogs anyone?

Branding. A company that provides data to remote programmers can request branding as a condition of the service.

Platform lock in. As Microsoft has demonstrated time and time again, a platform strategy beats an application strategy every time. Once you become part of the platform that other applications rely on, you are a key part of the computing infrastructure, and very difficult to dislodge. The companies that knowingly take their data assets and make them indispensable to developers will cement their role as a key part of the computing infrastructure.

Goodwill. Especially in the fast-moving high-tech industry, the “coolness” factor can make a huge difference both in attracting customers and in attracting the best staff.”

That doesn’t clearly translate into traditional business models necessarily, but if you look at key business breakthroughs in the past, the picture today becomes more clear.

1. The first breakthrough business model was based around page views. The domain created an Apple-like controlled container. Exposure to eyeballs was sold by the thousands per domain. All the software and content was owned and operated by the domain owner, except the user’s browser. All you needed was to get and keep eyeballs on your domain.
2. The second breakthrough business model emerged out of innovations in distribution. By building a powerful distribution center and direct connections with the user experience, advertising could be sold both where people began their online experiences and at the various independent domain stacks where they landed. Inventory beget spending beget redistribution beget inventory…it started to look a lot like network effects as it matured.
3. The third breakthrough business model seems to be a riff on its predecessors and looks less and less like an operating system. The next breakthrough is network effects.

Network effects happen when the value of the entire network increases with each node added to the network. The telephone is the classic example, where every telephone becomes more valuable with each new phone in the network.

This is in contrast to TVs which don’t care or even notice if more TVs plug in.

Recommendation engines are the ultimate network effect lubricator. The more people shop at Amazon, the better their recommendation engine gets…which, in turn, helps people buy more stuff at Amazon.

Network effects are built around unique and useful nodes with transparent and highly accessible connection points. Social networks are a good example because they use a person’s profile as a node and a person’s email address as a connection point.

Network effects can be built around other things like keyword-tagged URLs (del.icio.us), shared photos (flickr), songs played (last.fm), news items about locations (outside.in).

The contribution of each data point wherever that may happen makes the aggregate pool more valuable. And as long as there are obvious and open ways for those data points to talk to each other and other systems, then network effects are enabled.

Launching successful network effect businesses is no easy task. The value a participant can extract from the network must be higher than the cost of adding a node in the network. The network’s purpose and its output must be indespensible to the node creators.

Massively distributed network effects require some unique characteristics to form. Value not only has to build with each new node, but the value of each node needs to increase as it gets leveraged in other ways in the network.

For example, my email address has become an enabler around the Internet. Every site that requires a login is going to capture my email address. And as I build a relationship with those sites, my email address becomes increasingly important to me. Not only is having an email address adding value to the entire network of email addresses, but the value of my email address increases for me with each service that is able to leverage my investment in my email address.

Then the core services built around my email address start to increase in value, too.

For example, when I turned on my iPhone and discovered that my Yahoo! Address Book was automatically cooked right in without any manual importing, I suddenly realized that my Yahoo! Address Book has been a constant in my life ever since I got my first Yahoo! email address back in the ’90’s. I haven’t kept it current, but it has followed me from job to job in a way that Outlook has never been able to do.

My Yahoo! Address Book is becoming more and more valuable to me. And my iPhone is more compelling because of my investment in my email address and my address book.

Now, if the network was an operating system, there would be taxes to pay. Apple would have to pay a tax for accessing my address book, and I would have to pay a tax to keep my address book at Yahoo!. Nobody wins in that scenario.

User data needs to be open and accessible in meaningful ways, and revenue needs to be built as a result of the effects of having open data rather than as a margin-based cost-control business.

But Dare Obasanjo insightfully exposes the flaw in reducing openness around identity to individual control alone:

“One of the bitter truths about “Web 2.0″ is that your data isn’t all that interesting, our data on the other hand is very interesting…A lot of “Web 2.0″ websites provide value to their users via wisdom of the crowds appproaches such as tagging or recommendations which are simply not possible with a single user’s data set or with a small set of users.”

Clearly, one of the most successful revenue-driving opportunities in the networked economy is advertising. It makes sense that it would be since so many of the most powerful network effects are built on people’s profiles and their relationships with other people. No wonder advertisers can’t spend enough money online to reach their targets.

It will be interesting to see how some of the clever startups leveraging network effects such as Wesabe think about advertising.

Wesabe have built network effects around people’s spending behavior. As you track your finances and pull in your personal banking data, Wesabe makes loose connections between your transactions and other people who have made similar transactions. Each new person and each new transaction creates more value in the aggregate pool. You then discover other people who have advice about spending in ways that are highly relevant to you.

I’ve been a fan of Netflix for a long time now, but when Wesabe showed me that lots of Netflix customers were switching to Blockbuster, I had to investigate and before long decided to switch, too. Wesabe knew to advise me based on my purchasing behavior which is a much stronger indicator of my interests than my reading behavior.

Advertisers should be drooling at the prospects of reaching people on Wesabe. No doubt Netflix should encourage their loyal subscribers to use Wesabe, too.

The many explicit clues about my interests I leave around the Internet — my listening behavior at last.fm, my information needs I express in del.icio.us, my address book relationships, my purchasing behavior in Wesabe — are all incredibly fruitful data points that advertisers want access to.

And with managed distribution, a powerful ad platform could form around these explicit behaviors that can be loosely connected everywhere I go.

Netflix could automatically find me while I’m reading a movie review on a friend’s blog or even at The New York Times and offer me a discount to re-subscribe. I’m sure they would love to pay lots of money for an ad that was so precisely targeted.

That blogger and The New York Times would be happy share revenue back to the ad platform provider who enabled such precise targeting that resulted in higher payouts overall.

And I might actually come back to Netflix if I saw that ad. Who knows, I might even start paying more attention to ads if they started to find me rather than interrupt me.

This is why the Internet looks less and less like an operating system to me. Network effects look different to me in the way people participate in them and extract value from them, the way data and technologies connect to them, and the way markets and revenue streams build off of them.

Operating systems are about command-and-control distribution points, whereas network effects are about joining vectors to create leverage.

I know little about the mathematical nuances of chaos theory, but it offers some relevant philosophical approaches to understanding what network effects are about. Wikipedia addresses how chaos theory affects organizational development:

“Most of the focus on chaos theory is primarily rooted in the underlying patterns found in an otherwise chaotic enviornment, more specifically, concepts such as self-organization, bifurcation and self-similarity…

Self-organization, as opposed to natural or social selection, is a dynamic change within the organization where system changes are made by recalculating, re-inventing and modifying its structure in order to adapt, survive, grow and develop. Self-organization is the result of re-invention and creative adaptation due to the introduction of, or being in a constant state of, perturbed equilibrium.”

Yes, my PC is often in a state of ‘perturbed equilibrium’ but not because it wants to be.

The recent blog frenzy over hyperlocal media inspired me to have a look at Outside.in again.

It’s not just the high profile backers and the intense competitive set that make Outside.in worth a second look. There’s something very compelling in the way they are connecting data that seems like it matters.

My initial thought when it launched was that this idea had been done before too many times already. Topix.net appeared to be a dominant player in the local news space, not to mention similar but different kinds of local efforts at startups like Yelp and amongst all the big dotcoms.

And even from their strong position, Topix’s location-based news media aggregaton model was kind of, I don’t know, uninteresting. I’m not impressed with local media coverage these days, in general, so why would an aggregator of mediocre coverage be any more interesting than what I discover through my RSS reader?

But I think Outside.in starts to give some insight into how local media could be done right…how it could be more interesting and, more importantly, useful.

The light triggered for me when I read Jon Udell’s post on “the data finds the data”. He explains how data can be a vector through which otherwise unrelated people meet eachother, a theme that continues to resonate for me.

Media brands have traditionally been good at connecting the masses to eachother and to marketers. But the expectation of how directly people feel connected to other individuals by the media they share has changed.

Whereas the brand once provided a vector for connections, data has become the vehicle for people to meet people now. Zip code, for example, enables people to find people. So does marital status, date and time, school, music taste, work history. There are tons of data points that enable direct human-to-human discovery and interaction in ways that media brands could only accomplish in abstract ways in the past.

URLs can enable connections, too. Jon goes on to explain:

“On June 17 I bookmarked this item from Mike Caulfield… On June 19 I noticed that Jim Groom had responded to Mike’s post. Ten days later I noticed that Mike had become Jim’s new favorite blogger.

I don’t know whether Jim subscribes to my bookmark feed or not, but if he does, that would be the likely vector for this nice bit of manufactured serendipity. I’d been wanting to introduce Mike at KSC to Jim (and his innovative team) at UMW. It would be delightful to have accomplished that introduction by simply publishing a bookmark.”

Now, Outside.in allows me to post URLs much like one would do in Newsvine or Digg any number of other collaborative citizen media services. But Outside.in leverages the zip code data point as the topical vector rather than a set of predetermined one-size-fits-all categories. It then allows miscellaneous tagging to be the subservient navigational pivot.

Suddenly, I feel like I can have a real impact on the site if I submit something. If there’s anything near a critical mass of people in the 94107 zip code on Outside.in then it’s likely my neighbors will be influenced by my posts.

Fred Wilson of Union Square Ventures explains:

“They’ve built a platform that placebloggers can submit their content to. Their platform “tags” that content with a geocode — an address, zip code, or city — and that renders a new page for every location that has tagged content. If you visit outside.in/10010, you’ll find out what’s going on in the neigborhood around Union Square Ventures. If you visit outside.in/back_bay, you’ll see what’s going on in Boston’s Back Bay neighborhood.”

Again, the local online media model isn’t new. In fact, it’s old. CitySearch in the US and UpMyStreet in the UK proved years ago that a market does in fact exist in local media somehwere somehow, but the market always feels fragile and susceptible to ghost town syndrome.

Umair Haque explains why local is so hard:

“Why doesn’t Craigslist choose small towns? Because there isn’t enough liquidity in the market. Let me put that another way. In cities, there are enough buyers and sellers to make markets work – whether of used stuff, new stuff, events, etc, etc.

In smaller towns, there just isn’t enough supply or demand.”

If they commit to building essentially micro media brands based exclusively on location I suspect Outside.in will run itself into the ground spending money to establish critical mass in every neighborhood around the world.

Now that they have a nice micro media approach that seems to work they may need to start thinking about macro media. In order to reach the deep dark corners of the physical grid, they should connect people in larger contexts, too. Here’s an example of what I mean…

I’m remodeling the Potrero Hill shack we call a house right now. It’s all I talk about outside of work, actually. And I need to understand things like how to design a kitchen, ways to work through building permits, and who can supply materials and services locally for this job.

There must be kitchen design experts around the world I can learn from. Equally, I’m sure there is a guy around the corner from me who can give me some tips on local services. Will Architectural Digest or Home & Garden connect me to these different people? No. Will The San Francisco Chronicle connect us? No.

Craigslist won’t even connect us, because that site is so much about the transaction.

I need help both from people who can connect on my interest vector in addition to the more local geographic vector. Without fluid connections on both vectors, I’m no better off than I was with my handy RSS reader and my favorite search engine.

Looking at how they’ve decided to structure their data, it seems Outside.in could pull this off and connect my global affinities with my local activities pretty easily.

This post is way too long already (sorry), but it’s worth pointing out some of the other interesting things they’re doing if you care to read on.

Outside.in is also building automatic semantic links with the contributors’ own blogs. By including my zip code in a blog post, Outside.in automatically drinks up that post and adds it into the pool. They even re-tag my post with the correct geodata and offer GeoRSS feeds back out to the world.

Here are the instructions:

“Any piece of content that is tagged with a zip code will be assigned to the corresponding area within outside.in’s system. You can include the zip code as either a tag or a category, depending on your blogging platform.”

I love this.

30Boxes does something similar where I can tell it to collect my Upcoming data, and it automatically imports events as I tag them in Upcoming.

They are also recognizing local contributors and shining light on them with prominant links. I can see who the key bloggers are in my area and perhaps even get a sense of which ones matter, not just who posts the most. I’m guessing they will apply the “people who like this contributor also like this contributor” type of logic to personalize the experience for visitors at some point.

Now what gets me really excited is to think about the ad model that could happen in this environment of machine-driven semantic relationships.

If they can identify relevant blog posts from local contributors, then I’m sure they could identify local coupons from good sources of coupon feeds.

Let’s say I’m the national Ace Hardware marketing guy, and I publish a feed of coupons. I might be able to empower all my local Ace franchises and affiliates to publish their own coupons for their own areas and get highly relevant distribution on Outside.in. Or I could also run a national coupon feed with zip code tags cooked into each item.

To Umair’s point, that kind of marketing will only pay off in major metros where the markets are stronger.

To help address the inventory problem, Outside.in could then offer to sell ad inventory on their contributors’ web sites. As an Outside.in contributor, I would happily run Center Hardware coupons, my local Ace affiliate, on my blog posts that talk about my remodelling project if someone gave them to me in some automated way.

If they do something like this then they will be able to serve both the major metros and the smaller hot spots that you can never predict will grow. Plus, the incentives for the individuals in the smaller communities start feeding the wider ecosystem that lives on the Outside.in platform.

Outside.in would be pushing leverage out to the edge both in terms of participation as they already do and in terms of revenue generation, a fantastic combination of forces that few media companies have figured out, yet.

I realize there are lots of ‘what ifs’ in this assessment. The company has a lot of work to do before they breakthrough, and none of it is easy. The good news for them is that they have something pretty solid that works today despite a crowded market.

Regardless, knowing Fred Wilson, Esther Dyson, John Seely Brown and Steven Berlin Johnson are behind it, among others, no doubt they are going to be one to watch.