Oblomovka

2002-12-04»

Much more detailed info than the keynote is Trevor Marshall's talk on security and antenna design and how not to leak signals in directions you don't want. Or, alternately, given that Marshall designed a parabolic feed that secured a 125Km WiFi link using standard Cisco cards, how to leak in exactly the directions you do want.) I suspect the talk is similiar to the one he gave to the SoCal Wireless Users group, which is streamed on the Linux Public Broadcasting Network if you want to learn more.

He also has some ugly-ass plots of PCMCIA antenna signal strengths (below - the two colours are different polarisations). WiFi card signals are all over the place, which I found out to my cost when I was trying to write a triangulating utility a while back.

I'm blogging Sky Dayton's keynote at 802.11 planet. There's not been much that's new, but it's a nice overview of Sky's take on the market.

Sky's "ah-ha" moment was when he was in Aspen, and he saw three APs in his hotel room. Two turned out to be wireless ISPs - both of whom starting scrapping between each other to talk to him. Reminded him of early days of Earthlink.

Survey says: 97% of travelling businessmen would alter their plans to gravitate to high-speed access (high-speed access is more important to them than wireless access)

Ubiquity in laptops - Dell will be putting WiFi in their laptops by default, new iPaq will have WiFi built in.

He sees WiFi in cellphones, cars, gameboys. How low can the power go on these things, I wonder? There must be some physical limit. Maybe we just need more sensitive APs? Sorry, mind wandering - back to Sky.

Only 3,000 commercial hotspots in the US - about a million potential locations (212 conference centers, 3032 train stations, 5352 airports, 72,720 business centers, 202,600 gas stations, 480,298 restaurants and cafes, 1,111,300 retail stores. Wow. There are more airports than train stations in the US.)

But how do you get to ubiquity?

Dayton compares it to early days of ISPs ("Nobody knew who was their customer and who was their competition"). Back then, everybody tried to do everything - owning the wires, the network, and the brands. Eventually each company concentrated in one area - end users are AOL, MSN, networks are UUNET etc, wires are the telcos. (Hmmm. Has this happened in broadband yet?)

Dayton's division in the WiFi market is: Venues (Mariot, Hilton, Borders, etc). Hot Spot Operators: "(Wayport, T-Mobile, Surf and Sip, etc). Aggregators, who take the fragmented networks and provide cohesion: Boingo, GRIC, iPass. And brands: Boingo, T-Mobile, Earthlink, Sprint, Cingular, AT&T.

Big potential, says Sky, is in Hot Spot Operators. Two aggregators max. But he would say that...

He believes that no one HSO (Hot Spot Operator) will be able to deploy more than 10% of the total footprint - because WiFi's short range, low barrier to entry, and venue fragmentation. If your brother runs a chain of coffee shops, you could set up a HSO and roll it out, and steal past any bigger HSO. That's why aggregators are necessary.

Mac version of Boingo sniffer in Q1 2003. Not much talk about *how* it aggregates with HSO.

Some nattering about Hot Spot in a Box - allows any access point becomes a commercial hot spot. Costs $500 includes hardware, will drop to $300 as they talk to major manufacturers -- any broadband end point could become a hotspot. There must be a way of turning that into a software app. What kind of cut do these Boingo resellers get?

Hardware with Boingo built-in: Nomadix, Colubris, and Vernier. Fairly minor players. Client software carried with Orinoco, D-Link, SMC, Netgear, HP laptops, Earthlink (doh), Fiberlink. Yeah, but who pays attention to the CDs in the box?

Audience seem a bit underwhelmed. You can see that WiFi is taking off: lots of low-attention-span wide-eyed MBA wolverines sniffing and snarling their business plans on the floor. They already know Sky's overview - they want to know what everyone is going to do for them, or what they can do to everyone else.

The conference is much much bigger than last year, but I think the smarts has got a bit more dilute. Ah, well - once more into the J-curve...

Now, should find Glen and feed him some Theraflu?

A friend wrote asking me if a petabyte would be an affordable amount of disk storage in five years time. Blowed if I know, but in scrabbling to answer, I did find this great projection of the next twenty years of magnetic storage. It contains this graph which is based on IBM estimates of future storage capacity and price. (which I couldn't be bothered to convert into HTML - hey, but maybe you will and sent the text to me. Updated 2004-07-11: For my birthday, Adrian Furby sent me this HTMLised. Thanks, Adrian!)

As you can see, a petabyte will still cost about $70,000 in 2008. Consolation prize: a terabyte will sell for $70. Better start saving those files now!

According to a cursory IMDB search, Jason Schultz has discovered that 93% of the movies released from 1927-1946 are unavailable (it'd be interesting to include current TV showings in this - what was that site that let you grep through US TV listings for keywords?). As an interesting aside, the IMDB has 36,386 titles for that period. The copyright office says there are 37,144 - which means that the IMDB, a largely amateur effort, has snagged 98% of the titles. All goes to show that Kevin Kelly's assertion that enthusiasts might be better at preserving film history than paid copyright holders might turn out to be true. (Off of the rc3.org)

2002-12-03»

A new and better patch for the old linux-wlan on the 2.4 kernel; an addition to How To Wash Dishes that was meant to be added in March (sorry Dave.)

A lesson in game design, hosted on IRC.

How many inhabitants does Sunnydale need to support its apparently limitless supply of vampires? Brian Thomas, PhD candidate in ecology at Stanford, investigates. (From More Like This)

It's December, so it must be time for the Perl Advent Calendar: one CPAN module explained (with brief tutorial) every day. Day one taught me about finding URLs in text, day two explained how to write to files in place with automatic data recovery if it all goes wrong, and day three showed me a all-Perl Perl-and-C mini-SQL database in less than 300KB.

2002-12-02»

I use public transport a lot - even in Silicon Valley, where it sucketh like the Black Hole of Calcutta. To do so in such bus and train deadzone requires exactly the kind of juggling of schedules and careful dead-reckoning navigation that I am utterly lame at.

This is why I depend on online journey planners (like these versions for San Francisco and London). I also walk around quite a bit, trying hard not to get lost. A little pocket GPS and an online street address to latitude/longitude converter have revolutionised my wanderings, around and apart from public transport trips. I hardly ever get lost anymore, and I rarely underestimate how long it will take to get somewhere.

There's just a whole stack of destinations that this tech has let me see are perfectly reachable without a car, even in California.

Now add into the mix services like NextBus, which monitors and shares info on the realtime position of all the buses in San Francisco. Wrap it all into some portable device (or wireless service), that lets me provide an street address, and plots a route and ETA on the fly.

I feel confident with more realtime info, and realtime positioning, a lot more folk would be tempted by public transport, or indeed walking, than now.

That it takes slightly longer to get places doesn't bother me - I get a lot more done on a bus or train than I do even as a passenger in a car. The cost is a pain in the Valley ($4.00 for a day pass), but in most decently-run trans areas that burden is less acute.

No, the largest hurdle public transport has to overcome, I think, is the feeling of powerlessness and unpredictability it induces in most people. I think you can go a long way to reducing that - without requiring any heavy initial investments in public transport itself, by harnassing this new tech. It'll never be for anyone - but it's certainly increased my usage, and appreciation of, even one of the flimsiest public trans system in the world.

2002-12-01»

So I googled for a biography of Winston Churchill, and this was the second hit. It's a fine potted history, but ends on an odd note:

...During all of his life he had served no less than six British monarchs: Queen Victoria, Edward VII, George IV, Edward VIII, George VI and Elisabeth II.

He also possessed a large collection of toy soldiers. However in his book 'My Early Life' he does not mention which make of soldiers he collected. They were probably all made between 1880 and 1900 and therefore some time before Lineol and Hausser figures became available. As a collector he might be the only one of whom also a personality figure was produced.

... and then I realise. This is the biography of Winston Churchill, as it affects the world of toy soldiers. Did that make it any less useful, though?

2002-11-30»

SpinSanity is exactly the sort of site that I don't visit as much as I'd like, but would read every day in a RSS reader like NetNewsWire. And now they've added a RSS feed, I can. Hooray!

Except I'm not using OS X much these days, so I don't have an RSS reader. Time to play around with Straw.

Over seven hundred slides from Peter Gutman's Godzilla Crypto Tutorial. I imagine aliens flipping through this in about thirty seconds to absorb all current knowledge regarding the uses and abuses of crypto. Not particularly mathematically heavy-weight, but pretty techie in parts. The seventy-one slide guide to crypto politics is pretty comprehensive (not entirely up to date though). I got this from the enhyper financial crypto digital library mailing list, which occasionally turns up some great resources.

And I thought burning a video of the ultrasound onto CD-ROM was high-tech. This New Mexico firm is doing 3D ultrasound images, with a haptic interface so you can reach out (in?) and touch your unborn child. I admit to being mildly freaked.

2002-11-28»

One of the first cross-cultural fidgets I learnt in America: the English use "please" on more occasions than Americans do; OTOH Americans expect "thank-you" when English people wouldn't miss it at all (or would be deploying it sarcastically). So this is one way we manage to sound rude and overpolite to each other at the same time. So, thanks. Thanks!

Anyway: New get your war on. New reverse-link engine to see who is linking to your blog.

Rupert Goodwins did the BBC interview on Palladium I mentioned. He says that the Microsoft rep, Stuart Okin said that Pd was being pushed by Microsoft to become TCPA 1.2. That's to say, not complement or extend, but actively replace the old standard. Anyone heard this said in public before?

2002-11-27»

I'm still recovering from flu, which I caught in San Jose, just before flying to New York. I still have it now, here in Portland, and apparently I'm still contagious (I had to abandon going over to a nursing home to help Quinn's grandfather today for fear of spreading it). I got it from Gilbert, and I do wonder if he picked it up from the Boston LISA conference. My hosts in New York caught it while I was there, and I'm sure I must have handed it to at least a couple of the people at the Social Software summit, who will be displacing it all over the world.

I wonder what strain it was? It was the worse flu I've had for years, and snagged everybody in our house by the time it had finished. But the CDC report for this week (permanent link for November 16th flu report here) says that there's been no noticeable flu activity in California.

I wonder if that will change in the next few weeks. I wonder if the strain will spread in other areas, and I wonder that if it does whether I might have something to do with that. I feel like Typhoid Mary.

Less egotistically, Quinn asks how anarcho-capitalist systems without centralised disease control would deal with epidemiology: how do you introduce standards to monitor, control and eliminate, say, smallpox without a central organising force? I wonder if our amazement at vanquishing smallpox and TB by centrally setting standards isn't partly fuelled by a sense that epidemics are rarely controllable centrally at all: that these are the exceptions rather than the rules, and require massive acts of co-operation to work. Disease control is the act of a mature and very sophisticated centralised system: is it fair to compare it with simple models of decentralisation?

Doesn't answer the question, though: Q thinks that this is up there with National Defence as a life-without-the-state showstopper. It's one of those warning flags that makes most of us feel that radical decentralisation might tumble into something rather more horrid than the bright lights of Libertaria.