I remember, in Palo Alto during the Internet boom, everything was so intense, everyone could sense the historical forces, the people, and the money at work ... "it feels like a war," I quipped, "just massive birth instead of massive death."
But the only similarity I see now, between war and boom, is that I've become an Internet news junkie again. And a vertebra in my neck is hurting once more. The differences are more striking. For example, the difference between "Internet Time", as we first perceived it, and now.
Just try to watch how quickly important stories move around today ... it's enough to give you whiplash. Ah, perhaps that explains my neck pain.
I followed the Peter Arnett thing today ... he went from working for NBC and National Geographic, to the Daily Mirror, in an instant.
But even more amazing is how fast history can be rewritten. Arnett's Baghdad stories disappeared from NBC. I saw the entries in the search database, but they clicked through to nowhere.
Just got a note from Dan Gillmor, saying that he found Workspot bogglingly slow.
On the other hand, a real-time data guru, John Sahr, who uses the passive radar of commercial FM broadcasts to monitor the shape of the earth's ionosphere, says: "I logged in using Netscape on a RH7.3 laptop over a 384k DSL link, performance was amazingly good, considering what was happening. It was obviously slower than a "regular" link, but definitely fast enough to use. "
It's a good thing we have the demo, so people can self-select for speed! I wrote Dan that he may have a constrained connection, but "performance perception is a delicate issue ... it's possible that our subscribers tend to be intrigued by latency, rather than suffer from it ... "
During a raging anti-war party in downtown Eugene, a small group of shy technology people held a monthly Eugene Linux user's group meeting. Like all Eugene groups, the meeting was replete with characters far too complex to invent. This may be one of the few cities whose non-profit free software group, efn.org, has a storefront in the city's center. I'm tempted to open a workspot.org shop next to it.
The topic of the meeting: Knoppix, the Linux Terminal Server Project, and Workspot. Knoppix works extremely well, although it still helps to be a Linux admin, depending on what you need to do with it. LTSP is used for the Internet café at efn.org.
Bob Miller organized a working core group for the demo, with laptops and several extra machines and pieces of equipment. The group pieced together and debugged the demo in front of a live audience (albeit a small one). Like a live engineering project, Bob acting a the chief engineer. Actually much more interesting than trying to 'produce' a slick, foolproof demo. The process is as educational as the product.
From a dozen people, only one person was really interested in proselytizing the Linux desktop. He was only partly satisfied with Knoppix -- it's great, but still rather install-like. He was about to leave before the Workspot demo: I stopped him, and asked Bob to launch a browser on Knoppix, which I'd never done before, and go to workspot.com. Mozilla came up, luckily with Java pre-installed, and I said 'hit the button'! Voila, the desktop appears! With enough attention to robustness, sometimes things will work in the worst of circumstances.
Late night talk with Christopher Alexander ... I mention the one-click Linux demo. He says that should be a very big deal, to Linux people. I said it should be a big deal to half-a-billion people.
He looks at me with his 'be honest' look and says "Is Linux the best desktop?" I say, no, Mac OS X and Windows XP are still better in several ways, although no desktop is really going in a good overall direction. He said "you're not being a very good salesman. What's the point in showing people a Linux desktop that's not better than what they have?"
He said that the $100 cost of an operating system, amortized over the life of the machine, is very little money. I said that for big purchases by corporations and governments, it makes a difference. "Is that your market? Bureaucrats?"
No, certainly Workspot's for individuals. And individuals usually choose for themselves. So they need to see something better. Linux is popular among people who can use its strengths. Either people need to grow to be more Linux-like (which isn't too common, but it's kind of the goal of many LUGS); or, Linux has to work better for people. When it does, then we'll build real momentum on the desktop.
So, we must make a better desktop.
You know, about a week ago, I started to write a wild manifesto for Internet corporations -- about the need for openness and honesty; about gathering customers and partners by having good intentions and making them visible; about resisting the temptation to spin PR fantasies, and instead helping to reveal what really goes on in the world; about effective companies without the trappings of corporate offices; about working with your heart on your sleeve, instead of hiding behind the badge and business card of sterile professionalism; about the desperate need to move capital and commerce towards truly public-minded, sustainable, humane, personal and meaningful work ...
... and then I started my blog, and such ranting seemed redundant.
In the next week, I'll try to get other CEOs to blog. It's not hard: certainly they write enough office memos! It's about time they wrote memos to their public ...
***
Chris Gulker just publicly shared a moment of his enthusiasm as a Workspot user. He's a legitimate fan ... all his correspondence to me comes from his workspot.net e-mail address [using Ximian in a browser beats webmail anyday.]
If you Google Gulker, you find out about his secret past as a high-end Hollywood paparazzi. I like his photo of Brooke Shields. [He's now putting all his photos online! Excellent! He's photographed everyone: Gore Vidal, Lisa Lyons ... he even caught me in my old haunt in Palo Alto.]
***
Chris Gulker quoted me anonymously during the dotcom boom, when I described Workspot as fulfilling yet another computing dream: the immortal machine ...
What kind of dream is that? Here's a sample motivation for 'immortality':
Up until the mid-1990's, architect Christopher Alexander was happily using software under a long discarded DOS-based graphics system known as GEM. He was really fond of the software, it was simple and elegant for writing books, and honestly, he produced excellent work with it. "Why on earth would I stop using this?" he asked me.
Because of ... lots of bad reasons, basically. It was getting harder to incorporate the format into any other document, to translate it, to print it, to e-mail it ... etc. Everyone has a similar experience ... forced to abandon something good for something contemporary.
This isn't right! Lots of old software is wonderful! Thousands of projects on the web are dedicated to good old software. In fact, the obsession with Linux owes a huge debt to the resonance of the original brilliant papers on software tools, which came out of Bell Labs some 30 years ago.
Immortal machines? I have lots of old code, which I'd release under GPL, and start a project around ... if it only ran on something! FreeDOS, OpenVMS, BSD & Linux will make that possible in theory ...
... but on a practical basis, once something runs, I want it to continue to run! Is that asking so much? I know the reasons ... techno-cultural shifts, or whatever they're called at the moment.
But we have the opportunity now to fix this situation. We could keep these environments running on Workspot forever, for people to use whenever they need it.
A fellow wrote me from a Portland company, saying he'd like to put his software up on Workspot, to demo to some colleagues, but it only runs on RedHat 7.3 at the moment. Early this year, we updated all our machines to RH 8.0. So, although we have a better system, our immortal machine is already suffering bit-rot.
Virtual machines (see Thursday) can take care of that problem ... once they're up, probably in the summer, we'll just keep lots of Linux distros around ... we'll keep around as many versions as we can, and make sure they work. (We couldn't do that with Windows ...)
Most importantly, when we arrange old software in a showcase, and attach it to micropayments, programmers could be paid to support good, existing application software. Today, the typical software business model is: release worse software, with new addictive additives, as often as possible!
***
I'll long-jump past the boundary of bleeding-edge here, and say that an immortal machine needs more than versions of operating environments. It need to save running systems ...
If I have a desktop with a dozen windows open (or emacs with two dozen buffers) it's because I'm working on some specific thing, and I need programs, manuals and massive context. I want to take all that, and store it, for a year if I have to. I want to be able to revive it, to pass it on to someone else ... I want it to be a memory aid.
If this sounds like going overboard to be user friendly, look into the field of process migration. Many groups like this are freezing the state of an entire system.
This doesn't just let you leave a cluttered desktop for later ... it lets you clone a cluttered desktop so you can give it to someone else to use; it lets you show someone an unfolding, evolving system ...
... and it lets you checkpoint your work when it's in a good state, so you can experiment, screw it up and still go back to where it was working last! This is really critical for learning how to do good work ... it's hard to learn to draw properly without an eraser! [That may sound glib, but you can get into terrible habits if you teach yourself to draw, from life, with only a pen. I have a whole site of drawings serving witness to this!]
Because of the existence for decades of version systems, like CVS, many computer people realize this ...
But I remember a mutual-enthusiasm session I had with Will Wright last year, where we explained this exact thing to each other, in the context of Chris Alexander's generative sequences. We imagined a vast network of instruction, much like A Pattern Language, which users would navigate in order to build things. And then Chris came in, and we explained what we were talking about, and the cleverness of our infinite back-tracking idea, and he said, basically, that if you learn to work deeply, and "if a sequence is really good, then you won't have to correct much". Infinite support for user mistakes could lead to infinitely shallow users.
***
Originally, Workspot was interested in process migration so we could send a user's work to a nearby datacenter, to cut down on latency. This caused our crypto-colleague Efrem Lipkin [whose early work is chronicled in Steven Levy's book Hackers] to go into a hilarious and profound meditation, for weeks, on the subject of process migration to untrusted servers. Can you keep a process encrypted as it's running? Think about it.
But it turned out to be moot. Most of Workspot's latency develops in traveling the last mile, not the thousands of miles along the backbone. It works fine in Japan.
A simple thing, which people can't yet do with their Workspot, is install a program as root.
Another simple thing, which I want, is for people to run freshly contributed software on Workspot. Let's say that 'mad genius X' has just written a program that will change everything ... people should be able to run such freshmeat without installing it on their home machine.
But we can't yet set up a system where we encourage everyone to distribute standalone programs, and run them. If someone offers you an executable in their directory, it could plant something nasty in your directory, the moment you use it.
All this implies disposable machines: virtual machines on which you could install anything as root; and virtual machines (with a virtual filesystem, or journalling and rollback) on which you could run anything as a normal user.
Programs could go through the normal community acceptance promotions, until they were tested, read and trusted enough to run at home.
[Someday, of course, all home machines, which will run Linux, should have these features and handle these issues automatically].
When I started working on Workspot, in 1998, it was obvious that we'd have to tackle virtual machines. Bruce Robertson started writing a VM Manager, and watching the progress of other projects, even proprietary ones like VMWare (something we were wary of, because we'd have to change it so much). A founder of VMWare visited us in Palo Alto ... but nothing happened before the bubble burst.
A number of GPL VM initiatives that have emerged (I recommend Grant Gross's Newsforge summary):
The technical issues are interesting. At the moment our VNC server spawns sessions on a cluster of machines -- we'd either need it to spawn a session within a VM, or direct the user to remote login to that VM from their main Workspot desktop. The same choice applies to starting a fresh VM.
But which makes more sense from a user standpoint? There are many scenarios where we'll need to spawn a VM without initiating a desktop -- for example, when we run a single application in a web page ... (I'll write about this, Appspot, soon.) So we need to create HTML navigation through the Software Showcase anyway.
But someday, especially since we are desktop Linux advocates, we'll need to create a desktop interface to all this variety. Luckily, we have Nautilus, and it should be satisfactory to use directories to navigate, through all the possible machines and distros, cooked and raw software.
The main issue with the raw software is protecting the user's files. Since our first time around, journaling filesystems, which keep a transaction journal of filesystem writes, and allow you to rollback in case of trouble, have become more cooked. I'll write about these soon.
Hey, with virtual machines running, we'll be able to demonstrate the different kinds of Linux filesystems too! And work on sensible, humane user interfaces to filesystem rollback! One step at a time ...
We're in banner rotation at DesktopLinux.com, as part of our referral program.
The best part of partnering with these folks, is that they'll help me fulfill some computing dreams.
Workspot should be a Showcase in Cyberspace (catchy? kitschy?) for all Linux software and hardware.
The very existence of this showcase should improve what's produced. I'd like the software to change in quick cycles, in front of people's eyes ... programming as performance art!
This is a pretty tall order, but we should be able to organize it by the summer. With everyone watching and helping, improbable things still happen in Internet time!
All of these vendors and projects, large and small, need money. One of the things we'll have is a micropayment system -- software rental for pennies.
Imagine you need to use some crazy specialized high-end workstation application. Today, you can't usually try it without contacting a salesman. So most people won't try it! That's no good for anyone.
But what if you could try it for a few cents, and then compare it to the competition, for a few cents more? To me, this meets a straightforward consumer need, to test drive and sample, to see the breadth of possibilities. It's like a real marketplace; a tangible Consumer Guide. Users can sample for as long as they want -- they're being charged as they go. Vendors get money, feedback, and a growing client base. Everyone wins.
It's amazing to me that this isn't available yet. I mentioned it publicly years ago, and still it hasn't happened. We'll do it here, and we'll do it first for Free Software.
There's hardware I'd like to see too ... I think many people would like to testdrive servers from Sun, HP and IBM. Embedded chip emulators could be very useful. And why not a small Beowulf cluster? ... or Deep Blue Jr. ? Or film rendering engines? Supercomputing for the masses!
*
Jeremy White, who founded the Wine-leveraging company CodeWeavers, is interim chair of the Desktop Linux Consortium. He's put me on the advisory board. Heh heh ... GNU/Linux will push Microsoft off the desktop, blotter and all. It's inevitable.
Dear diary, so much happened today.
The demo's up! Anyone in the world with an Internet connection can now try a GNU/Linux desktop instantly -- with two clicks (including this one)! Ha! Another first!
Oops, I should've make it one-click! Some advocate for humans I am!
Hey, one-click Linux -- that would be another first! (Yawn). What the heck, let's do it:
I guess I'm still shy from when the Amazon patent for one-click purchase came through. We were doing the Palm wireless app for Barnes & Noble, and the court order came down, and B&N had to change everything on their site immediately ... I don't know exactly what happened, but Curt wandered over to me, shaking his head, and said the frantic bn.com chief engineer just disgorged the entire system's code at him, "Here! Figure it out yourself!"
Speaking of Amazon patents, we have a referral program in place, redolent of this Amazon patent: "an Internet-based referral system that enables individuals and other business entities ('associates') to market products, in return for a commission, that are sold from a merchant's website". I'm not the only one that finds it incredible that they received a patent on reselling over the Internet. Why not just patent capitalism? Or conspicuous consumption? Then everyone would have a good excuse to get out of the rat race.
Well, we need to do this referral program anyway. Perhaps someone at KP will invest ten dollars in Workspot, so we don't get a court order? No. That's the coward's way out ... instead, I want every Linux site to do Workspot referrals ... they'll make money this way (and they need it!) and if Amazon's lawyers call, the riot will shake the Internet to its roots, and we can organize a massive overthrow of all this nonsense.
Our first two referral volunteers to face the lawyers:
... the weblog of Chris Gulker. He writes for one of my favorite British papers, The Independent.
... and Bruce Sterling's Viridian Design site: a Green celebration of the end of post-modernism. (I need to get him to read The Nature of Order ...oh -- Bruce: "Yeah, Chris Alexander, he's an interesting guy. My architect made us read PATTERN LANGUAGE before building this house that I'm sitting in.")
Another hot story: our VNC applet now tells you what ports it's trying to use to connect to the Linux desktop. Cool. Could even be a diagnostic tool. When we're sure it's working well, the code changes will go on workspot.org.
Had a great talk with Jill, the desktop dynamo from DesktopLinux.com. With the energy of people like her, in five years nobody will be using Microsoft on the desktop. [Oops! I hope I didn't just put her life in danger!]
I want to quote from my original letter to her:
"There are perhaps a half billion people, with Internet
connections, who don't run Linux on their desktop. The point
of DesktopLinux.com, as it comes across to me, is to (1) convert
these masses and (2) organize the revolutionary
working group.
These goals also drive Workspot. It's a GNU/Linux desktop
demonstration, for anyone with an Internet connection.
And it's a place where a specific desktop can be used and
seen simultaneously by many people. With pre-installed
examples, it can facilitate discussion of the human qualities
of features and applications. It can speed up and broaden
the adoption of good structure and function. It could be a
platform where developers get instant feedback from a large
audience ...
I know it can do these things, because we do it everyday.
Our happy users grab our attention daily, and on a daily
basis, we fix things, and move to the next issue. This is
the way desktops should be improved, using a quick&small
cycle, an incremental approach, uncovering the largest problems
with an active test group.
This is the strength of the Internet community, and yet it's
barely tapped in the application and desktop platform world.
Workspot offers this. All I need is community agreement: that Linux needs an online demo. "
I've always said that Workspot is an open corporation. But where's the proof?
Does any tech ceo do a blog yet? (Besides Dave Winer, of course). Please write me and right me, if I'm wrong. I know Alan Meckler is trying this, but he's more of a media guy. I'm pure geek.
Secrecy gets you nowhere in life. People who hide ideas tend to stop having them. I think Jerry Weinberg wrote something like that. [Even if you give away ideas as fast as he does, you should still be cited, when possible.]
In business, the hardest thing, and the most important thing, is mingling with the real world. So secrecy only hurts. Customers and partners and employees and friends all benefit from openness and transparency. If everyone really knows what you're doing: potential competitors can become partners; potential victims can stop you; mutual support can flourish. Competition? Sheesh. The purpose of business should be to make the world better, so there's plenty of work to go around!
But, of course, beware: a blog like this only makes it seem like I'm telling you everything.
Years ago, I think I was with Jay Walljasper at Utne Reader's magazine-rack-centered offices in Minneapolis, and we started talking about those "Correction" columns in newspapers. And how such a column implies that everything in the paper is true, that they left no stone unturned, except for a few minor oversights! [Chomsky or Cockburn probably pointed this out first].
But even if I'm honest and thorough, a blog like this can only offer glimpses. I can't show the whole. Well, I can try. Ok, I won't try, I'll do it. [Hm. No. I'll avoid the Yoda citation.]