Joel on Longhorn

Joel on Software has a length article on the sea change at Microsoft that is culminating with Longhorn. Joel was a programmer at Microsoft but now does server side programming and thinks that MS is losing their API monopoly not another company but to the web. A brief synopsis:

• Backwards compatibility of Window Applications is being sacrificed for wholesale new technologies (.NET, WinFX) that require developers to scrap everything they have done before. To use some aspects of Longhorn (Winforms, .NET remoting) will require rewriting applications which were rewritten for .NET just a few years ago.
• Most new developers find writing Browser based applications much easier and a better return on investment
• Most users a fine with using Browser based applications (I kind of disagree with this one - browser based applictations, particularily the ones which really should be client / server or desktop applications tend to be horrible, unusable hacks. OTOH, people do use them, usually because they are forced to as part of an enterprise solution...)
• In the 1980 / 90s the number of computers was growing at such a high rate that new computers were the rule rather than the exception because new computer sales were larger than the installed base). In this environment, developers could afford to target new machines / OS but now you can't get away with that because many people are still using Windows 98. How many people have the .NET runtime or can run the .NET runtime? (Again browser-based software is not a panacea here - many web sites are pretty choosy about which browser they will work with - usually the aforementioned "enterprise applications" but installing a current browser is seen as less of an issue than installing an OS / OS Component)

For Joel, it seems like upgrading to Longhorn as a development platform is not a compelling deal in terms of what can be done versus the cost of doing things. The Mac is a good case study for this - how many people have switched to OS X? I'm fairly sure it's not a high number of total mac users (although Apple has told developers that "we don't care and you shouldn't either). There is another issue though because as an ISV you don't care about the total # of users but the total # who are going to buy software at all - in that case the people who upgrade OS are also the people more likely to buy your software so the numbers are skewed somewhat. But upgrading Windows is hard. Really hard. So I think it's far to say there are significant #s of people who are several revs behind on Windows OS who still buy software.