I'm guessing you've probably never sent a greeting card with a Strawberry Shortcake on it. The company that makes those, American Greetings, sold $1.5 billion dollars worth of product last year. Much of it was paper, although they also have an online division which charges a lot of people a lot of money to send birthday cards. (I only know this because my mom says she pays $50 a year for the animated flash cards she mails me all the time.)
What relevance does this have for desktop software? Simple: it should remind you that there is an entire universe of users out there who do not have Paul Graham's problems. Many of these users have problems which can be solved best by desktop software. Those users will keep desktop software companies in business for a while. (Did I mention one point five billion dollars of paper greeting cards? And this is HOW many years after email?)
As long as I'm on the subject, ubiquitous net connectivity gives us desktop developers some interesting options against piracy. One is moving value from the client to the server. I do it a little these days (for non-piracy reasons), and if I can figure it out I'm sure the rest of the industry can.
You can't pirate a desktop app that requires connecting to a server to function (+) any more than you can pirate Basecamp by downloading Firefox.
+ Edit for clarity: I don't mean in the sense of "phone home", I mean in the sense of "value is provided by code or data which only exists on the server". Think WoW: without connectivity its the world's most expensively engineered paperweight.
That's why the adapted Wilde quote applies. I didn't say desktop sw will instantly disappear, just inexorably.
I have six online competitors to my desktop application that I know about. They're made by wonderful people. Some of them are even free.
Many of them show ads for my product and they're, empirically, the best places for me to put ads.
(Google will helpfully let you see what websites are making you money. Other AdWords users should really get to be friends with the Placement Report. Fascinating stuff there. Tangent over.)
Given the choice, today, between a no-install, free online application which is staring them in the face and clicking to another website, downloading something, installing it, running it, playing around, clicking purchase, reaching for the credit card, typing in their details, and inputing a license key, a few hundred people have said "Oh, easy -- I want the downloaded one."
Why is that? There are a couple of reasons (which you would be intensely bored about because you've presumably never had to create bingo cards for 25 people), but it boils down to "The experience of using the desktop application sucks a lot less than the experience of using the web application, for at least some users". They get something out of the desktop version that they couldn't get from the web version, and voted with their wallet.
That's just the view from my little, teeny wedge of the economy. Are all similar wedges going to inexorably vanish? Probably, if you come at it from the assumption that everyone will eventually come to work like you. To misquote another Englishman: "There are more needs in heaven and earth, Horatio, than are answerable by the types of products you yourself buy."
I bet a fair share of desktop software persists soley because of how awkward it is to print from the browser. (The reason no one realizes that the URL, page number, and timestamp can be removed is entirely because of the unnecessary complexity of the UI.)
If Mozilla, Google, or Apple made their printing interface dramatically better (which would be synonymous with making it dramatically simpler) they might be able to win a surprising number of IE users. The correlation between technical proficiency and a penchant for hardcopies seems roughly inversely proportional. While, for instance, my parents don't see much additional value in using Chrome over IE for browsing, they'd probably use it exclusively if it made printing significantly more comfortable.
See? One step ahead!