Postcards from the Bleeding Edge
Thursday, August 07, 2003
Just a couple hundred million folks, sitting around talking
 | Pericat's Unlocking the Air throws off some sparks from the hypersphere with with a thoughtful piece on the nature of Internet conversation, called "Wanna Talk"?
|
The last number I remember for the size of the internet community was about 100 million back in 1999 or so.
It's potentially over 6 billion times over, if you just count the people living today.
Or, if you want to use metcalfe's law for "the utility or usefulness of a network equals the square of the number of users"...
Internet usefulness[1999] = 100,000,000^2 = 10,000,000,000,000,000
I don't have current numbers for the size of the net (anyone?) - assuming 500 million -
Internet usefulness[2003] = 500,000,000^2 = 250,000,000,000,000,000
How useful is a usefulness number? Beats me. All these connections are only useful in potentia - I'm going to research this a little bit in light of Dan Power's LinuxWorld speech on grid computing, (the grid has 2 million computers connected to it now!) and see what comparisons I can come up with. Connections in the human brain? In ASCI White? Complexity of the genome? Etc. (anyone got numbers for these?)
Pericat mentions Ursula Franklin's The Real World of Technology (I have to read this) and also points to Burningbird's musings on the semantic web. I've been trying to talk coherently about what people like Berners-Lee are doing nowadays, using phrases like "conceptualize concepts" and "talking about talking". Now I don't have to, I can point them at BurningBird instead.
Good reading. (btw, happy b-day, pericat!)
¶ 6:26 PM 0 comments
Book publishers at LinuxWorld... and VOIP
Prentice-Hall had galleys of Eric Raymond's new book: The Art of Unix Programming. I've just spent a few hours browsing it online, rather than finishing my LinuxWorld summary. I like the detailed comparisons between the design methods of various open source projects. Given the wealth of historical data now openly available I wish more researchers/writers/managers had thought to examine that - and come up with "Why"'s, not "How"s. Eric also spends much time talking about design patterns. Brook's book, "The Mythical Man-Month", has influenced the computer industry for 30 years, perhaps Eric's will do for the next 10. My favorite part is the collection of "rootless root" koans...If I'd read The Career Programmer: Guerilla Tactics for an Imperfect World 6 years ago, I might have had a lot more fun. And I'd be working now. The book's from a publisher I'd never heard of before.
O'Reilly had an absolute ton of new books out, but the one that caught my eye was an older one on VOIP using VOCAL. I saw 5 people browsing it in the booth at the same time.
Consider all the VOIP enabled PBX's out there. Quicknet's been doing Linux Telephony for years. There's multiple software stacks, open source ones like Openh323, Vovida, and asterix, and commercial ones from multiple big-name vendors. There's multiple products from Cisco, and Sprint....
Notable at the show: Michael Robertson announced his new Sipphone - IP handset. 129.99 for the phone. Or 250 for two.
Who needs local telephony when what you need is an extra line out to a friend across town? I still remember as a kid using radios to talk between houses late at night when I thought my parents were asleep... think of that teenager you have that rings up big bills on the cell and hogs the landline you have that costs 40/month, and encourage him/her to get an IP phone instead...
Got DSL and a network in your house or business, why not (at least) run your internal phone calls over it?
The basic infrastructure exists for Voice over IP telephony to finally rise above the radar. Will it? Yes. This year? I don't know. ¶ 3:48 PM 0 comments
Tuesday, August 05, 2003
Live from LinuxWorld - w/Jim Gettys
This year I got a press pass. Love that free food in the press room!I went from the booth of a vendor (to remain nameless) that had things like TinyX wallbatising their boards. I played dumb and asked the salesdroid: "What do you mean by X?"
He replied "We do a lot of work with the Intel Xscale processor. We do design, initial bring up, everything you need in the design cycle to ship your product in 8-9 weeks, tops. We've got multimedia/video/networking all..."
Me: "So that doesn't mean Xwindows..."
He: "Oh, yea, we do Xwindows too."
He kicked back into the memorized hardware salespitch again... it was obvious he'd ingested too much coffee prior to facing the LinuxWorld mob, so I moved on. I do Xwindows.
I hooked up (briefly) with one of the godfathers of Xwindows graphics system, Jim Gettys himself. He hails from HP and currently gets his kicks from hacking on arm based pdas via the site handhelds.org. HP's booth, right by the door, showed off HP's ipaq series running two versions of Linux. There was a full Debian release, running off a local flash hard drive, and Familiar, a version that fits (comfortably) in less than 32MB of flash. On familiar were the current releases of Opie and GPE - the competing open source environments for handheld Linux PCs.
I had a chance to play with the GPE release. It's NICE. It's pretty. It's functional. I gotta get home and download it for my ipaq - it's GTK 2.2 based... But anyway, I'm here to meet people, not hardware... and I caught Jim's attention.
| Taht | Has anyone commercialized handhelds.org's technology? |
| Gettys | Various people are doing stuff with it or are ready to deploy it, there's some security stuff is about to be deployed, for example. It's not to say that I believe that Linux is ready to be a volume product in place of PocketPC. I think that will take a few more years. It's the vertical things that Linux can do that WinCE and and PalmOS can't touch.
it's going to take 1-2 more years til we catch up, but after then we are playing to our strengths rather than theirs. |
| Taht | Lately I've been playing games with where metcalfe's and moore's law intersect. One thing I noticed is that flash ram costs less than half as much this year as last year. |
| Gettys | I understand. That's why the people doing framebuffer based stuff are all wet. With another turn of the crank it won't be that much to have regular qt and regular GTK on the same Ipaq. In fact, with most of our current models (not the low end ones) that's true already. |
| Taht | I don't think that Moore's law dominates the cost of these devices. There's Gene's law which describes power consumption, then there's the cost of the actual display itself. |
| Gettys | Different technologies are on a different technology curve. Disk drives traditionally were at a very low rate of 20-30% per year, and then for reasons I don't fully understand they suddenly 5-7 years ago went to a faster than semi-conductor improvement rate. |
| Taht | How about the cost of displays... |
| Gettys | I don't know. If I did, I couldn't tell you. |
| Taht | Anything else going on? |
| Gettys | A bunch of us (the X.org folks, the freedesktop.org folks) are trying to get a large scale engineering project focused the fundamental technologies required on the desktop. It's a larger charter than just the X server itself. X is just a piece of what we need. It's not funding we're looking for, but talent. |
The XFree86 project forked a while back, with the Xwin group of developers splitting off. The philosophical differences between the design paradigms between the XFree group and the Xwin Group are readily apparent via the difference in their website designs....
The news today is that the infrastructure for the fork is almost ready...
From today's discussion on xwin.org...
- Xouvert (pronounced zoo-vehrt) is a development branch of
XFree86 provided to try experimental features and
enhancements, as well as changes to the XFree86 architecture.
We hope that XFree86 will use our code once we have
somehting to contribute.
We will use arch (arch.fifthvision.net) and the code should be
available very very soon (within a day or two). We'll be using
the savannah service (savannah.gnu.org). The project is
registered already, the domain names (xouvert.org,
xouvert.com, xouvert.net) are registered. Pretty soon (probably
once the arch repository is available) I will submit my patch for
full backingstore support (not just for windows that request it)
and other enhancements are in the pipeline. We intend for the
project to be as open as possible.
In 1998, Jim wrote an excellent article on The Two Edged sword - on his reconversion to X. He reconverted me with it. It's taken far longer than I ever dreamed it would for X to re-become viable - but I look at the quality of redhat's bluecurve, gnome 2, and kde 3, gtk2 and qt 3, and then back at 1998, when none of that stuff existed - and I think that the prospects for Linux on the desktop in the next 5 years look bright indeed.
I'm going to go catch a session on clustering samba and then the keynote.
¶ 2:59 PM 0 comments
