Re: HA philosphy

["David Gumbrell" <david.gumbrell@xxxxxxxxx>]

If you want to understand what living in an adaptive smarthouse would
be like and you've ever owned a TiVo then consider the logical
extension of those weird program recommendations that you can't
understand how it is arriving at, and no matter how much you
"punish"
it by hitting the Red Thumb, can't seem to get rid of.

Call me a luddite, but I am unconvinced that adaptive (AI) approaches
are going to be acceptable. The more sensors, the more exceptional the
cases leading to strange behaviour, but also the harder to unwind and
remove. I don't think people would accept losing control of their
environment: "Computer says no".

Instead I expect a pair of new paradigms to emerge, one for the
operation of smarthomes and one for programming (or behavioural
definitions).

Currently pretty much everyone over a certain age understands mains
switches, partly by learning and partly by convention - ie light
switches tend to be higher than power switches and placed near doors.
But I bet that this paradigm was arrived at after much experimentation
on the introduction of electric light.

This reminds me of trying to explain the use of the green and red
telephone buttons on a mobile phone to a perfectly adept user of
traditional telephony. And even the green and red took time to settle
out, with early mobiles trying all kinds of symbolics.

So it is reasonable to expect new paradigms for the operation of a
SmartHome. Lest people reject the chances of these being acceptable, I
am reminded of Steve Jobs' response when challenged that people would
have to learn to type to use his computers: "Time will solve that
problem".

As for programming, I recently read a study which attempted to develop
a simple means to predict which computer science students were going
to be successful and which were not. What was interesting was the
large percentage who *simply could not grasp programming* (at least
with conventional programming languages) as well as the simple test
that was developed and, within a limited study size, appeared to be a
good predictor.

So I take it as a given that behavioural definitions are going to have
to be achieved with other means than classical programming languages.
Pre-canned functionality may work - "good enough" can be
sufficient to
be commercially successful, although us tweakers would probably reject
such a system.

One technique would be to build on other paradigms, for example
labelled softkeys on phones might be a good model for the primary
operational interface, although the current "one device - one
control"
model (mains switches) would not survive. Similarly, a behavioural
definition UI might build on spreadsheet and database grid type
approaches.

But I hope that there may be far better ideas out there, and I am sure
we will see much experimentation as Moore's Law makes it cheaper to
have richer UI devices.

Cheers,

David


** Sponsored by http://www.BERBLE.com **
all the Cool Stuff, in one Place