Internet 3.0
Monday, February 19, 2001 by Dave Winer.
Good morning sports fans! 
A week of schmoozing is good for idea flow. Coming back from the P2P conference, I can see that things have changed. The bright eyes are back.
2001 will be the most pivotal year in our industry since 1994. What we do now will set the direction of the technology industry for the next ten or twenty years, as the 1994 corner-turn did.
In 1994 we moved to centralize. Now we're moving to decentralized systems, for good economic reasons. The money to pay for centralized services has evaporated. In the next iteration of the Internet we'll create thinner servers and thicker clients.
This realignment will change the balance in the software business. Sun, whose tremendous growth in the 90s was derived at least partially from the push to centralized networking, will have to form links with software on the desktop. It was a good move for them to acquire StarOffice and Cobalt, and they probably should make some kind of deal with Apple, the major non-Microsoft desktop system vendor.
Like it or not, the tide is turning to favor Microsoft now. They had to struggle to make sense in the thick-server model of the 90s. Their products are already designed to work well in a decentralized model. Perhaps it's time for Microsoft to change its profile, to turn from the scrappy competitor to a leader-with-stature. There are signs pointing in both directions.
Cycles repeat
Round and round the loop.
First there's a technological breakthrough, in 1994 the networking stack reached the user interface. Before the Web it required a good imagination to see the benefits of networking. In the Web it was in your face, with Dancing Hamsters and Elian and Bill Clinton, Slashdot and News.Com, weblogs and Salon, eBay, AOL and Yahoo. As a result we now have a fantastic networking infrastructure that we didn't have before 1994.
As the personal computer moved the cursor out of the mainframe "glass palaces," the next version of the Internet will move the software and data that's now on centralized servers to the users' desktops, and it will evolve from there.
In talks last week at the P2P conference I found it useful to call this Internet 3.0, to focus on the software that until now has remained invisible to all but a few. It's there, and you're going to be running some of it, soon. (And it won't just come from Microsoft.)
Internet 3.0
Version 1 was the pre-Web Internet, the playground of techies and geeks and professors and programmers. Gopher, FTP, email, newsgroups.
Version 2 was the Web, instant messaging and email. Broad adoption. You can buy movie tickets on the Web. Internet kiosks and cafes are everywhere. URLs on all business cards. Who needs the Yellow Pages when we have Yahoo?
Internet 3.0 will realize the groupware vision of the late 80s which was really Doug Engelbart's vision of the 60s and 70s. Shared writing spaces with good boundaries. Structures that link to each other but are capable of managing greater complexity than the page-oriented metaphor of the Web. (Which few people read, everyone skims, so why not create interfaces that optimize for skimming.)
How will money flow in version 3.0? First we need easy-to-program financial services available through SOAP and XML-RPC. A bank implemented in software. New companies will form, ones that leverage the communication tools to an advantage, to create products that are uniquely tuned to customers. Internet 2.0 brought us online car purchases, eBay, bill-paying, banking. Those were profound changes. Version 3.0 will refine this by giving us better tools for working with the new power.
Internet 3.0 will be a bootstrap. No steps will be skipped. This is why static low-tech formats based on XML are central to the next level of services, alongside the dynamic SOAP and XML-RPC-based services. It will scale from day one.
Internet 3.0 will not be dominated by Sun, Microsoft or a startup coming from left field. Contrary to what some believe, there will be lots of Dot-Nets. My small company, UserLand, will compete with Microsoft for the runtime of Internet 3.0, and for authoring tools. We will beat them, and they will beat us, in other words they'll have customers and so will we. If we can conceive doing this, so can you.
The bottom is open in version 3.0, as it was in 1.0 and 2.0, you can build off my success and have it flow through your software, and vice versa. The barriers are low, by design. You won't have to learn anything strange to latch onto the bus and get the flow to go through your bits. Same with Microsoft's software. You have my word on that. I'm watching carefully as we continue to work with them. If I see any attempt to lock in developers I will let you know. Same with Sun, IBM or anyone else.
The premise of all versions of the Internet is openness and choice. Even if it sometimes didn't appear that way while we were developing version 2.0, it was always true. We weren't locked in. Perhaps we can avoid this argument altogether in 3.0. (The only real threat is patents.)
As with Version 2.0 it won't matter if you use Unix, Macintosh or Windows. You'll just be able to do more on all those operating systems.
In Version 2.0 your performance monitor got flatter and flatter as CPU performance traveled up Moore's curve, but the architecture failed to take advantage of it. In Version 3.0 your CPU will do more work and your local hard disk will contain (at least) backup copies of the information stored on centralized servers. There will be more opportunities to write scripts that customize your desktop, and include your own creations in the user interface. You won't have to be an eyeball (sometimes it's relaxing), your brain will be more active in Internet 3.0.
Dave Winer