Three constraints that are holding the web back

Try using existing Java/Browser technology.

I would suggest this:

Each compatible web page has a signed applet which can also run resident on the remote machine via Webstart. If the server is not listening yet, start it.

If it is listening, send it a message. The Webstart remote applet then displays the message in the GUI.

My server just bound to a range of well known sockets, and the clients tried each one in turn. Try to keep it simple. I just wrote a string with a carriage return to the socket. No XML, no heaviweight protocol.

* Native Widgets - tough to get right

As soon as you start having native widgets, you cross over into having the users preferences honoured. Great for user, hard for site-consistency and support. Joel Spolksy recommends a complex setup[2] , just to get it right.

This may be a little challenging for the average webapp. Admirable but tough.

* Payments

Good suggestions from the article. A bigger problem I see is the large payment processing fees levied by each different payment provider

For instance £0.20 per transaction when payment is via Debit Cards in UK. If you're selling small items, that can be 35% of the price!!

---

[1] These days I use it to tell me when my ant builds have finished

[2] http://www.joelonsoftware.com/articles/fog0000000306.html

2) wince I used to design UI for desktop apps. Menu items have been a dead concept for quite some time, and with good reason. They disconnect the user from the task or object at hand. There's no obvious mental mapping between a menu item and what's on the screen. The web is better off without these.

3) Tipjoy.com =)

- Start a listener in Firefox

- Get alerts from websites in the bookmarks toolbar

- Display subtle alert count next to icon in bookmarks toolbar.

No disruption.

Users can control who can send them alerts.

Simply add/remove icons from bookmarks toolbar.

1. Desktop Notifications: Terrific idea, and going to take a longass time to happen in a unified way. Already getting implemented via XMPP, perhaps browsers will just end up including an XMPP client where you can add to the roster from the DOM (pending confirmation). If this ends up happening via Twitter instead, there is no hope for this world.

2. Native Menus: a thoroughly mediocre idea -- for starters your webapps should not have fucking desktop-style menus! Furthermore how would you implement File-Edit-View on a mac? If you really need a dropdown, <select> is right fucking there. The only redeeming native-ui feature I can think of would be the ability to add to the browser's context menu, but that would be ripe for abuse if it was even close to useful enough.

3. MICROPAYMENTS: The single shittiest idea in the history of the web, and it's never going to fucking go away. Fucklers will continue to tilt at this windmill forever, and will always fail. There is far more terrific content available that I want to peruse than I will ever have time for, and I'm a guy that spends every waking moment reading. I'm serious, I often read well over 1M words a week (though I haven't instrumented in over a year), almost exclusively on the web, though in college I got 100+ books a year through ILL. There is just too much out there to be nickel & dimed.