Timing is everything

A friend IM'd me this morning exasperated at a co-worker who looks at bugs found during QA as being the same as bugs found in earlier testing. Traditional metrics* show that the cost of fixing bugs found early in development is much less than the cost of fixing bugs found late -- like during Q A.

Keep that in mind as you read the memo from Bill Gates that was Slashdotted today. The memo is shown on Gizmodo and is pages and pages of Gates explaining what is wrong with an experience of using a Microsoft product. He describes a lengthy experience that begins when he just wants to download some software and is told that they didn't expect anyone to want to download the software from the download page. So he has to download a ton of software through the update mechanism and reboot his machine only to find that he still doesn't have the software he wanted in the first place.

Towards the end of Gates' memo he writes "The lack of attention to usability represented by these experiences blows my mind."

The article ends with a note that Gates is asked about this memo and reports that he wrote something like this every day. That part is good -- that's part of Gates' role at Microsoft -- or at least it was.

So I was thinking about one of the last MacHack's. A member of the iPod team told stories about the development of the device. He said that Steve Jobs took the prototype home each weekend and on Monday they got a bunch of notes on what was still wrong with the device. Jobs caught little things. Things most of us would never notice. The engineer described one bug where the time elapsed and the time remaining would not always add up to the same number. There were times when they didn't quite move in sync. Jobs noticed and it was fixed.

So although I know better than to generalize from single data points, I'm going to. Both leaders are paying attention to details and the user experience in important ways. One key difference is that Gates is catching problems with software that's already shipping while Jobs is catching problems with a device before it ships.

=====
* This is the moral equivalent of "it can be shown" and means that my first cursory glance through some books didn't turn up the actual studies with the accompanying numbers. I think the curve is exponential with fixing bugs found at development time being about one tenth the cost of those found during QA which are in turn one tenth the cost of those found after the software is released. OTOH I could just be making those figures up.