Google Speaker Event – Adam Bosworth

The second event on Monday was the first in 2007 Google Speaker events. This looks to be the start of a series of events that Google is planning in their new space in New York. And a good, large space it is, even if painted a depressing battleship grey. But the lovely spread of food and the free beer or wine made up for that 😉

The first speaker was Adam Bosworth, talking about Physics, Speed and Psychology. Although as he explained it the title was designed to try and get the organiser to cancel the talk! He initially thought that the target audience was going to be a small group of Googlers and had written it up that way, but did not change it when told it was to a far larger (about 150?) external group.

Nothing new or ground breaking here, Adam just talked through some of his history, trying to put explanations into why things failed or succeeded. Overall, extremely enjoyable, espeically about the history of Ajax, even if he felt he was moving occasionally into the area his girlfriend had described as POF territory, ie pontificating old fart.

In 96/97 Adam was working on a team that was developing DHTML, know known as AJAX. As far as they determined, the reasons for moving this way were sound = it was strategically well thought out as it looked like the internet was moving applications to a thin client, not the thick client beloved by Microsoft at the time (so the team was not exaclty the most popular with their message) and people were going to need a high level of interactivity on the web, similar to the current Windows apps. So they went ahead and built this office package, with spreadsheets and presentations and word processors but no one used it. Their assumptions were completely wrong.

The companies they were trying to sell this to did not like it (Adam paraphased their reaction as ‘stop developing this stuff, go away, we hate you). The rich functionality of the apps meant that it would need a high level of support, of which very few of the customers had. In 1997, web apps were used occasionally, not day day out as office aps were. Therefore the barrier to learning it in a useful way meant that everything needed to be extremely simple to reduce the need for support; it needed to be intuituve so that if you only visited it a few times a month or less, you could still work it. The other barrier was speed; to build any of these apps in JS takes a lot of JS, a lot of bandwidth and the connections, and chips, were too slow. The delay and unpredictability of the response meant it was never liked.

Unsurprisingly, anything over .5sec reaction produces frustration; the best reaction time is less than this but it appears that people can live with half a second. Variations in network speed also meant that the reactions of the apps were extremely unpredictable, the system did not work consistently so turned people off very quickly.

Ten years later though and Ajax is well into its second life. The physics has got better, bandwidth is far, far faster and so are the chips. Carefully crafted applications are fast enough for people to use (although if you’ve tried Yahoo TV guide you can see it’s easy to build something that is a step backwards in functionality. The other psychological change is that people are using web apps far more; the increase in use frequency means that the interfaces can be richer.

Next up for examination was the PDAs, which have gone through a similar life cycle. The first iterations were complicated, pen computing, requiring writing recognition which was unpredictable and did not work very well. The second iteration carried on with the writing, but asked that the human learned to write in a special language. Whilst still unpredictable and slow, it was an improvement. The third iteration, the Blackberry and Treo just decided to go with the keyboard. You can work faster, it’s predictable, there’s no translation delay between input and it appearing in the screen.

The slowness and lag is why mobile browsing has been slow on the uptake. SMS, the simple text message, took of by accident. It was built for testing but with it’s simplicity, speed, asynchronicity helped spread usage. Now physics has caught up with the vision on mobile computing and browsing on a mobile device is practical and not too painful.

The final area looked at was natural language, which has failed so many times. If you can ‘talk’ to a computer in your natural language you expect it to behave like a human and demand a precision that is no there. But it’s resurrection came in part from Microsoft Help, which needed some way to help people navigate it’s vast depository of help documents. Search, with it’s fuzzy logic, is perfect for natural language, it works when it doesn’t really because it is a time saver, it filters the noise out there.

So the key lessons Adam has learnt?

  • Think about people’s activities
  • Determine the frequency of use; the less it it used the more simple it needs to be.
  • If it takes greater than 2 seconds to perform a task, break it down to smaller chunks – don;t make people wait

A fun talk, full of anecdotes and some useful advice. Google are planning to do a fair few events from the sound of it, so plenty to look forward to.

This work is licensed under a Creative Commons Attribution-NonCommercial 4.0 International License.

Comments are closed.