Agreed. There are many things still not possible on the web.
Office/productivity apps? I'd say Google Docs is a far cry from the performance of MS Office or OpenOffice.
Likewise, 3d performance on the web is miserable at best. You might get a bit of help from Papervision or Java's 3d engine, but it's not suitable for large, interactive, 3d applications.
How about copying and pasting from one web app to another? :-)
I love web apps, don't get me wrong, but I'm agreeing that there are many times when they just won't work.
"... How about copying and pasting from one web app to another? :-) ..."
The one I hate is "right mouse click" usability. There is no standard so each app has a different approach. Others don't bother and the RH mouse click is for the browser totally wrecking the illusion of an application.
There is a game called Wurm Online, made in Java, that you might want to take a look. It might change your opinion about 3D. It is not very polished, but that's more the devs' fault than the platform itself.
Why Web-Apps? Ease of use, ubiquity, no installation.
Why not Java-Webstart? Because Java is not cool enough.
Why not Flex/Silverlight? Hold on, that might be the next hype.
Fad always plays an important role in main-stream utilisation of technology!
"But today there is Java (write once, run anywhere) and Java Web Start (auto-upgrade to all your client machines). [...] You can use traditional client/server, with rich clients written in Java and deployed by Java Web Start talking directly to your database using JPA/Hibernate/JDBC. [...] But even then you can use Java and Java Web Start, and your client application can talk to the middle/web tier using a service API."
Cappucino delivers an entire application in JavaScript. It doesn't use HTML or CSS or any of that stuff. When I say "web app" I assume an HTML-based web app. Flex and Silverlight and even Java are also "web frameworks" if you mean that they talk with the server over the Internet...
Doesn't have to be Java. That's just what I use. It's not evangelization, just a preference. Java is a distraction, my point is more about architecture than any particular manifestation of the architecture.
I suspect that we are too focused on web applications as potential businesses to the exclusion of other ideas.
I recently attended a "Startup Camp" barcamp thing. We had an idea session, and it started with the reading of a prepared list of ideas -- the audience stopped that, and requested first that the list be filtered to only software ideas; then when a desktop app idea was read out, requested the list be filtered to "server software", and when one of the ideas was a rsync type backup system for your servers, they further explained that by "server software" they meant only web apps.
I think this is mainly because web apps are perceived as requiring little capital to launch, compared to other ideas, especially hardware. Desktop software is (rightly) suspected of being a morass of weird APIs and hard work.
Java and Java-Webstart are not the correct ways to write desktop software, in my opinion. If you only know Java and your application is mostly a one-off or demo, it will probably get the job done. But it is not encouraging to see most American programmers, when faced with the task of writing a computer program to run on the desktop, immediately get off in the weeds exploring various ways to make web apps run on the desktop. There is a large body research on how do to that, going back 30 years now, and lots of good ways, and systems to make it easy -- it's sad to see a supposedly technical person fiddle with trying to port web apps instead of just cranking out a program.
But back to the narrowness of current startup ideas: the cost of doing a hardware product is often over estimated. Don't knock it till you have checked out mfgquote.com. One of the major trends in hardware has been making things customizeable, such as FPGAs instead of ASICs, so no matter what you think up, it is probably not as hard as you immagine.
A common objection to hardware ideas is that "I will put a lot of capital into the idea and then cheap chinese manufacturers will copy it and rip me off." This is silly because webapps and software are even easier to copy.
The hardware idea that triggered the barcamp audience to request only web apps, was to take the "Mr. Lee catcam" and commercialize it. Unknown to people there (including myself), the guy who did that hack was already selling it -- check out http://www.mr-lee-catcam.de/pe_cc_o1_en.htm
I also think that Java Web Start is a pretty good solution when you have control of the computers where it is going to run and you need dynamic behavior. But I saw the same problem, people wanted a webapp.
Because web apps are on the web. Things that are on the web can be indexed. Things that are indexed can be found. Things that get found get users and make money.
Office/productivity apps? I'd say Google Docs is a far cry from the performance of MS Office or OpenOffice.
Likewise, 3d performance on the web is miserable at best. You might get a bit of help from Papervision or Java's 3d engine, but it's not suitable for large, interactive, 3d applications.
How about copying and pasting from one web app to another? :-)
I love web apps, don't get me wrong, but I'm agreeing that there are many times when they just won't work.