Ben Rosen is Back
Monday, April 26, 1999 by Dave Winer.
Last week independent web columnist Jakob Nielsen predicted that we'll be building websites primarily for pre-5.0 browsers until late 2003.
The message is unmistakeable. The rapid growth in functionality that rose around HTML and HTTP, the web, is over. Now the changes come slowly because the market has become slow at adopting change. There are so many reasons for it, everywhere you look, from any angle, the development of new features has stagnated, the new features that *are* being delivered are boring, there have been serious reliability problems in new versions, and when this happens, the market settles down, people upgrade slowly and carefully. The cost of breaking systems is too high because too much has been invested in working around the bugs in the installed base, each upgrade requires more careful transitions.
I also agree with Nielsen that there's a silver lining to this, web developers can use the slower pace to fill in new content management and serverside features such as searching and navigation directories and easier authoring. But don't miss this, as history teaches us, this kind of stagnation opens the door for a breakthrough, which just as inevitably follows the stagnation. It's a constant stream of flip-flops. Moore's Law keeps driving, and new generations of developers learn about things that were tossed out in the last generational flip.
We've seen this many times. The Apple II became a cluttered mess of upgrades and add-ons, which begat the relatively wide open spaces of the IBM PC. I remember thinking that 640K of RAM was enough for eternity. Four years later Lotus 1-2-3 had filled the memory, and the PC floundered with incompatible approaches to overcoming the memory limit.
Along came the Mac, with its linear address space and a non-Intel processor, it took hold, and it was time to convert character-based software to the graphic UI that came with the Mac, and later, Windows. Many leading products, including WordPerfect, dBASE and 1-2-3 were left behind in this transition.
Then in 1993, the web caught on and rewrote the rules. The graphic UI was thrown out and then hasitly and unsuccessfully added back (Java). Individual operating systems took a back seat, and HTML replaced the more expressive screen display technologies on Mac/Windows.
With the web, we took a big step forward, networking became part of the user interface, but we also took a big step backward as lineto-moveto was replaced with HTML's more clumsy <blockquote> and <table>.
So, in this context, Nielsen has, in my opinion, devised the math that proves that we're about to make another of these transitions.
So if the web browser is Visicalc, what will its Lotus 1-2-3 look like?
One easy bet would be to take another look at the GUI desktop and look for creative ways to connect that to the net. In fact, Microsoft is doing that in Windows 2000, in the integration with WebDAV; while not a complete emulation of the local file system, it's close.
But I think there's more promise in navigating from the opposite direction. Assuming that something like the web browser is the future platform for readers and writers, how would you navigate, from a software development standpoint, to something that did to the current web what Lotus did to Visicalc, or what the Mac did to PC-DOS?
Interestingly, net-aware games, notably Quake, and chat clients such as ICQ are paving the way. As they build out and people create SimCity-like browsers for network-based desktops, we won't need what W3C is developing. Just as the file format for 1-2-3 became a standard, the formats used by these apps to transmit their information will become the standards of the next decade. If Yahoo isn't running the servers behind these networks, or AOL or Microsoft or Excite, CNET, TheGlobe, Tripod or Xoom or whoever, they aren't going to fully participate in the next wave of growth.
Musical chairs, a great name and a great game. Six kids, five chairs. The music starts. Walk around the circle of chairs. When the music stops sit down on a chair. If you can't find one, you lose!
The same thing will happen in high-flow portal sites. It's an easy extrapolation that web browsers will be tied to specific websites in the future. We see this happening at Netscape (now owned by AOL), and for sure Microsoft will follow suit. Will Yahoo buy Opera? Looking five years out, I think it's a no-brainer that consolidation will occur around sites that also control their browsing technology.
Now you may not like this, you may think they all have to support the W3C standards, but they don't. Standards bodies can't make the kind of transition that's sure to come in networking in the next few years. To me, the activity around W3C looks like the stuff that goes on while the real transitions are happening in the market. The fights over expanded versus extended memory. Various mail APIs that were being pitted against each other while SMTP and POP were taking over the desktop mail systems.
Exciting software is what drives transitions. The IBM PC didn't go anywhere until 1-2-3 came out. And the Mac didn't boom until they erased the artificial 128K limit making way for Pagemaker, Photoshop, Excel, Director and other graphic apps that did something fun and useful with the memory.
Periods of openness are followed by periods of closedness. To assume that servers will be licensable, that the protocols will be discoverable, is naive. The software business is played nasty. Did you ever try to reverse-engineer the Excel or Word file formats? No matter what the standards-zealots say, the high ground belongs to the one who provides the software that excites users. In retrospect, Netscape blew it bigtime by yielding to the hype of the standards proponents. (Is it any surprise that Microsoft lead the parade?) Netscape could have, and should have, locked Microsoft out. Too bad they didn't have an architect with deep experience with software industry nastyness. They were led by a lamb and the sheep. It didn't work, at least the first time around. (AOL gets the nasty side of the business.)
The one that gets the best bet down, and manages to promote a client app that has greater appeal than their own server, will be the high-riser of the next decade. As 1-2-3 was derived from Visicalc, as the Mac had much in common with the PC, the next generation will look a lot like a web browser, but it will also look a lot *better* than web browsers, if you get what I mean.
Nielsen says it's going to be quiet in browser-land for the forseeable future. Who gets to make the noise that fills the silence? If history and human nature are any guide, that's where the next Mike Markkula, Ben Rosen or Jim Clark will place their bets.
I'm thinking about Ben Rosen these days. He's back in charge at Compaq. Rosen is a smart guy. I met him first in the early 80s when I was associated with Personal Software, the company that produced Visicalc. Rosen was an analyst and newsletter writer and smalltime investor. He was watching and listening carefully, waiting for his opening, and when the IBM PC came, he pounced, starting Lotus with Mitch Kapor to challenge Personal Software, and starting Compaq to provide an alternative to IBM.
I had a psychic Ben Rosen experience last week. Walking down a street in suburban Washington DC, I had a thought. I wonder what Ben Rosen is doing now. A couple of days later I got my answer on the front pages of the Times and the Journal. He's shaking up Compaq, where he's still Chairman, which is no longer just a PC maker. Compaq is DEC, Tandem, AltaVista, and the original desktop business. Rosen's creativity should be quite a force. I'd love to chat with him now to hear how he's thinking. Like I said, he's a smart creative guy.
Flipping it around, if I had Rosen's ear, here's what I would say.
"Get thee into software." (Traveling in the east has brought out my inner-Quaker.)
There's never been a better time to have strong plays in clients, workstations, servers and portals, and Compaq has strong positions in clients and servers and portals. What's missing is an offering for creative professionals, people who write, draw, animate, design and develop systems, in other words, developers. They surely sell the hardware that such people can use. But they can, in ways others can't, tie all those systems together.
We've gotten a big lift from having industry standard software, but that direction has run its course. Now I think it's time for experimentation, without the oversight of the standards bodies. Entertain creative people with more options. The path is totally clear, just add back the power the web took away. There are lots of excellent starting points. Macromedia's Flash comes to mind. Work with game developers and encourage them to link their interfaces up to productivity aggregators and connectors running on Unix servers. Encourage Apple to port QuickDraw and the various user interface APIs on the Mac OS.
What about Microsoft? NT didn't displace Unix, and some believe it didn't even displace Novell. They have no rabbit to chase in web browsers. Can Microsoft still strongarm the cloners? I don't know, but Rosen would. Microsoft still does the best job of responding to developers, but they haven't covered all the territory they need to. It's a good time to challenge Microsoft. There's room.
The open space is user experience. Cast a thousand seeds and let the developers lead. Microsoft probably will never really do this. It's the option that Apple had and failed to grab. But Rosen knows the history. He saw first-hand how the cart could be upset, now can he do it again?
A door can open, maybe a couple of them. The first door is the migration path for Win32 apps to Linux. The second door opens Linux to Macintosh API-based apps. This is a door we want to walk through. This path should be well-groomed, I wish it were, but it's not; I bet 1000 other developers feel the same way.
Linux needs a GUI, so does the web. But this is the small picture. The big picture, the software industry needs leadership. The conventional wisdom that we don't need software is incorrect, in fact it's always been software that new directions are formed from. Right now it looks like portals drive the industry, but if you believe that portals will leverage ownership of client and content development standards, it'll all come back to software eventually.
Another lesson of transitions is that the old way persists for a very long time. The 4.0-level browsers will be with us for the forseeable future. They will be the broadest least-common-denominator platform. There's still a lot of room for deepening tools and work environments for the existing web, just as there was new ground to explore in the Mac after Windows 3.0 shipped. Transitions require steady hands. Be ready to move, but also be prepared to continue in the same direction.