So what caused this problem? We released a storage software update that introduced the unexpected bug, which caused 0.02% of Gmail users to temporarily lose access to their email. When we discovered the problem, we immediately stopped the deployment of the new software and reverted to the old version.

source: Gmail Blog

150,000 users woke up to an empty inbox because they were part of a testing sample. This is routine for large production sites – they continually roll new features out to testing audiences. If Facebook has a new feature, they’ll put it in front of a test audience. As it is, I believe the storage (gmail offline) feature is an opt-in labs option.

It’s great that Google has a backup plan for a catastrophic data collapse. Hopefully they also have backups for the backup of their multiple location redundant data distribution network.

Obviously, Google does not mention anything about user testing or being more careful with its user tests. They are, however, very sorry. This is not to rail against user testing, it is much better that only 0.02% of the population was exposed to this malignant code.

Remember your backups. Only expose beta users to potentially dangerous code. Hey, they signed up for it.

[youtube]http://www.youtube.com/watch?v=5nt3gE9dGHQ&feature=player_embedded[/youtube]

This service works based on emails that a user reads and replies to most often. Priority Inbox learns a user’s preferences and delivers their interests. It is part of a coming wave of intelligent machines. Google, for example, has started scanning books, not necessarily to be read by people, but by machines to create a more knowledgeable machine.

This is happening now, and I believe there is an inherit flaw in the way prediction software is thought about. Although, I am no expert. It comes down to the idea that calibrating results based on preference may induce an echo chamber effect, helping lead to a more insular society.

That is to say, the algorithm seems to be based on what the user believes is important, but what if the user’s preference is wrong? Not in a ‘their ideas are wrong,’ but maybe in the sense that a person can never know what they truly need. Or, maybe they never realized a sphere of knowledge existed (very common in our diversity of people) . For example, my father has recently been talking about block time management and being a ‘slave to email.’ It is interesting to think about if these personalized algorithms will reinforce bad habits and behavior or serve as a gentle nudge in the other direction.

Netflix, for example, has been holding a competition to improve its recommendations for movies (source). While the algorithm is designed to give an oddball every now and then, we see that this methodology could quickly become an echo chamber. It’s nice to have a computer’s help, but you have to wonder at what point the machine is thinking for you.

I believe it comes down to the principle that a machine is only as good as its operator.

In other news: Starting in January, I will be starting a business based on my existing freelance career. I will probably do something different with this site, although it is on the back burner at the moment. Technical walk troughs have been the most popular, so one of my goals is to increase the coverage of that here.

There are lots of great science lectures but also business, education and politics all have freely available videos.

You can view the iTunes page for Stanford through this link: http://itunes.stanford.edu/

Might bring this thing back to life.

If you can’t read, apparently Facebook tells us “What doesn’t kill a quail only makes it stronger.”

Designed in the early 1990s by Mark Rosheim, the Omni-Hand is dexterous, rugged and hand-powered by an electric gearbox in the palm. It also was the most life-like and reliable hand that NASA made in the ’90s. The space agency’s researchers even put a glove on it. Courtesy of Wired