I think the problem is really that people see it as a binary issue. Browsers are extremely good for tasks where ease of access is important but they tend to fall down for tasks which require very flexible interfaces, fast response time or complex controls - the browsers simply don’t provide advanced programming constructs and, if you’re considering writing lots of ActiveX controls, the security / compatibility issues can drive the time up past the time needed to build a standalone application (this line can blur because you can do things like use those same ActiveX controls in a standalone app or embed a browser in it). I’ve noticed that some of Microsoft’s products started embedding IE for interfaces with mixed results - in some cases the net result is a less-efficient but prettier applet which was probably easier to develop.

A good example is content authoring - something like Microsoft Word is much better than browser-based editing for working on large documents; for small ones (e.g. weblogs), the differences may not be significant.

Compare also web based email to standalone email clients. Web-based clients are great for many things, but they tend to fall down once you need to do complex things - I like to search thousands of old messages for info from people or rapidly filter/sort mailboxes.

The real answer is to have both. I have my own IMAP server, which I can access using a webmail interface and a full IMAP client (including from wireless devices). The webmail interface is awesome for just quickly checking email but I still have the full capabilities I need for more complicated tasks.

Using your example of eBay - I think they could successfully market a full application for their hardcore users - picture all sorts of complex search tools, easy ways to track 25 auctions at once (multiple windows & panes), agent-type notifications, etc. Most of their customers wouldn’t use it but I think many of their heaviest users would. This doesn’t make sense to start a company (you want to reach as many as possible and the web-interface is good enough) but it is an interesting way to set your product apart in a mature market.

This argument is an artifact of woefully poor connectivity. It becomes a moot point when there is sufficient bandwidth to the last mile.

http://pdlogan.manilasites.com

So sometime in the 22nd century, we’ll all be OK?

Seriously, though, I can’t see how good bandwidth solves platform dependence or OS-version dependence. Good bandwidth lets my application run on Mac and Windows?

Financing bandwidth for everyone is a hard problem. I wasn’t claiming it would be easy, just that we need it. And once we have it, arguing about a 3270 interface vs. a Mac 512 interface will be moot. Here in the US I’d like to see a federal program for solving the bandwidth problem as an infrastructure problem. The government financed the highway system and integrated circuit technology as theywere considered to be in the public interest. The same should be done for bandwidth now.

As for platform independence… I am assuming that’s not much of a problem in the long run. Of course, bandwidth is indirectly related to that, at best.

I agree that we need better bandwidth, particularly out to the home - I like the idea of the government doing bounties or something similar to encourage R&D.

Platform independence is going to be around for a good while - I doubt different platforms will ever disappear, although we probably will have varying layers of emulation and virtualization. Some of the ways of accomplishing it (virtual machines, open standards) are also extremely important for making transparent networking secure - Java has most of the right ideas but Sun hasn’t provided all of the necessary framework yet.