I’ve just been reflecting on the failings of Wi-Fi. Boy, are there a lot of them. But what makes them interesting is the common root cause.
Problem #1: Wi-Fi isn’t sociable. There’s nothing in the protocol to announce who owns an access point. We don’t need some detailed, fixed protocol that fails to capture the market’s requirements. Just a URL (which might be hosted on the access point) where we can fetch a flexible XML document whose format might evolve. SSID isn’t good enough; we need meta-data too. I want neighbours to know what door to knock on to ask if they can have access!
Problem #2: Not enough ‘A’s. The heart of most mobile networks is the AAA (“triple A”) system, that authenticates, authorises and accounts. (Your acronym may vary. Don’t write, please.) Now there are independent standards like 802.1x for authentication that are all well and good. But Wi-Fi is lousy at authorisation and accounting. There’s no formal means of bootstrapping the payment process, negotiating your way onto the net. Fudged splash screens are OK until you start deploying non-browser applications like Wi-Fi phones. I don’t just want to distinguish between open and locked access points; I want to know which “open” ones are free.
Problem #3: It’s terrible to use. Channel selection. SSID selection. Encryption method selection. Litmus test: does your mum buy and install it, or does it need to wait for a Chrismas family get-together and offspring to perform system adminstration? You know the answer. Its usability sucks like a Dyson.
Problem #4: No partitioning. Everyone’s a blood relative, or everyone’s an enemy; network engineers don’t have friends. You either have totally open access, or lock-down. There’s no means of offering 2nd class access to passers-by, reserving primacy for the owner. No means of requesting registration of visitors, or recognising returning visitors.
So it’s kind of an autistic technology. It knows how to simulate Ethernet in the same way someone with autism might be able to memorise and draw a scene with incredible accuracy. But ask it to understand context, empathise and adapt, and it’s useless.
This isn’t an accident. The nerdy focus that makes an excellent wireless networking protocol engineer probably makes a lousy social scientist. The outcome of such engineering technical committees are solutions to engineering technical problems, not social or economic ones. It’s a bit like the (discredited) ontogeny recapitulates phylology motto from biology; technology development recapitulates standards committee composition.
Here’s a thought. Make each IEEE committee include some extra members: an economist, an anthropologist, a usability engineer, and a grandmother. The economist’s job is to look for scarce resources and help enable markets around them. The anthropologist considers how digital-era savages fashion tools to further social group interaction. The usability engineer examines the affordances of the technology; too few and it becomes inflexible, but too many render it unuseable.
Oh, and the grandmother? She’s there to remind everyone it’s all about people, not technology.
Posted by Martin Geddes at 08:12 PMTrackBack URL for this entry:
http://www.telepocalypse.net/cgi-sys/cgiwrap/mgeddes/MT/mt-tb.cgi/388.