There's some sensationalism in the editorialising around this piece, but I find it hard to argue with the core of what Gates is saying.
You're not going to be able to sell a piece of software based on the idea that it has less bugs. People expect software to be bug-free. When people buy a new version of a product, people expect new features. I mean, history has proven him completely correct - today, vendors provide free updates in their software to fix bugs and sell new versions with more features.
Secondly, you're always going to have bugs. It's inevitable. What's important is that your software, both in terms of design and testing, is resilient enough to avoid major bugs that a large proportion of people will run into. Historically, Microsoft were good at this - there's a reason they were behind "Writing Solid Code". What remains are bugs which are quirks and corner cases, which are generally less important and are harder to test for. Of course, the situation with security is rather different, as certain quirks and corner cases can have serious unintended repercussions.
I think it's worth making a distinction between "buying" a new version and "obtaining" a new version (where "buying" includes software that is charged for, even if you pirate it and don't pay for it). That is, the difference is between proprietary software and open source software.
When I pay money for software, I expect new features. When I go to download open source software, 9 times out of 10 I want the specific bug I'm dealing with to be fixed (and often it is). Heh, and often the bug I want fixed is that the software is lacking a certain feature -- but I've seen a lot of "issues" and "bug reports" listing feature requests, so the feature gets treated like a bug. Proprietary software is marketed in such a way that the latest version is all that you'll ever need, rarely are its deficiencies mentioned, and they definitely won't be mentioned as actionable "bugs".
The "versions" you're comparing aren't the same thing. Windows has a new version in the open source sense every month, and it's distributed automatically.
Is there any software company out there that ships products on the scale of Windows and Office and always fixes all known bugs in its code before RTM? If you know a bug will only affect 0.1% of users 0.1% of the time, it doesn't do anything showstopping even then, and fixing it will delay your release, you'd probably decide not to fix it too.
Disclaimer: I am working at MS this summer. But I would have said the same thing months ago.
Heck, if and when I find bugs that affect .1% of users .1% of the time, I'd put it at a lower priority than the release, even with a tiny project. Cost to benefit.
This shouldn't come as a surprise to anybody. Any nontrivial product has a bug database with hundreds of known bugs per module/component. Most are not worth fixing because regression testing is so expensive. And the expectation of end users is really low, people expect software to be lousy.
People even except phones to crash nowadays. In a Palm Pre review I read the other day they praised the Pre's stability: it "only" crashed 4 times during testing, and it was because "we were running a lot of apps at the same time".
Interview and transcript are not exactly the same. The transcript sounds more like, well, a transcript, the published interview is a lot smoother. Since all printed interviews are highly artificial products this is certainly to be expected. The transcript really seems to be the raw transcript. Even if it isn’t – the content is the same, the printed interview only sounds nicer.
I don’t think so. The linked transcript seems to be the real deal, the raw transcript of (a part of) the interview that was later translated for the magazine.
The transcript sounds pretty natural (then again, I’m not a native English speaker) even if it might not fit the usual style of Gates. The German interview sounds a little strange (you immediately notice that it was translated). I don’t think it was translated by a very skilled translator. I guess the English version would sound quite a bit more strange if you translated the German interview again.
Text for this page is extracted from the RISKS archive: <http://catless.ncl.ac.uk/Risks/17.44.html>;
This is the raw interview transcript (from which the magazine article was transcribed in German) kindly provided by the interviewer, Dr. Jürgen Scriba. The introductory text at the top is from Klaus Brunnstein, as found in <http://catless.ncl.ac.uk/Risks/17.43.html>.
I'm sure he wouldn't talk that way in a controlled presentation, but I believe it's likely that he could have answered that way in an interview back in 1995 in part because of his nature and the reasoning behind the statements.
I thought this was going to be a new interview, but I remember reading this interview and discussing it at work back in the 90s. I know, this anecdote doesn't really give this much of an air of legitimacy, but archive.org does confirm that it's been around that long (which of course validates my memory more so than if Bill Gates actually said this stuff).
He'd been in business for an awfully long time already in 1995. Where's the evidence to suggest he was so ridiculously unpolished after all those years?
Nobody says "No! No! No!" to an interviewer over and over again.
Oh please. A google search of "bill gates arrogance" returns over 200,000 hits. People like Bob Cringley and James Wallace have made small fortunes selling books about Gates's foibles, frustrations and outbursts.
He has a point though. I think if they said "you have to buy the next version to get the bug fixes", customers would be quite upset. Some companies do that, but really, these days I would expect to get bug fixes for free.
Even if the interview is fake, it's an interesting juxtaposition of business and technical perspectives: "Gates" is saying "people won't pay for bugfixes; but the interviewer is hearing it as "people don't want bugfixes".
I think bugfixes clearly have value, but it's a little unclear where exactly they are monetized. Consider: under what circumstances would existing bugs stop you from buying the present version?
- a 100% compatible and 100% feature-comparable alternative product... which additionally has fewer bugs.
- if you know about the bug before buying it. Perhaps this is the crucial factor: you only discover the specific bugs that are a problem for you some time after purchasing it, so they are not well-placed to be purchase-factor.
Bill Gates just owned the guy ! Amazing ... the interviewer tried to push the guy but he knows exactly what to do.
From my view, Mr Bill and Microsoft cant stop evolving because of the bugs. They are right on focusing in new features rather than on bugs. The problem is building to complex software. That´s the mistake, on my opinion. Sometimes i use office 2000 to write some document and i don´t feel any difference from the 2007 ... well ... both do the job. So why i need the 2007 ? Because it´s have more features ... just this. So the interviewer should go for that, in my opinion, "Mr Bill, why you guys don´t build more simple and less complex software ?" So Bill would probably answer what EVERYBODY KNOWS BUT STILL WANT TO LISTEN TO FROM HIM TO CRUSH THE GUY ... "Well, because simple software don´t give so much money"
My take (just from the provided page and not from further context) is that Gates was saying MS does not do a new version of software on the basis of / for the sole or primary purpose of fixing bugs. This does not mean they do not fix bugs in a new version. It means fixing bugs is NOT the motivation for doing a new version. Bug fixes do not provide sufficient motivation for users to purchase the new version.
EDIT: I did not mean for my comment to be a child to this parent. Clicked wrong when beginning it.
You're not going to be able to sell a piece of software based on the idea that it has less bugs. People expect software to be bug-free. When people buy a new version of a product, people expect new features. I mean, history has proven him completely correct - today, vendors provide free updates in their software to fix bugs and sell new versions with more features.
Secondly, you're always going to have bugs. It's inevitable. What's important is that your software, both in terms of design and testing, is resilient enough to avoid major bugs that a large proportion of people will run into. Historically, Microsoft were good at this - there's a reason they were behind "Writing Solid Code". What remains are bugs which are quirks and corner cases, which are generally less important and are harder to test for. Of course, the situation with security is rather different, as certain quirks and corner cases can have serious unintended repercussions.