If your app is highly data-driven, the default answer for what programming model to use is now "web application", barring special circumstances (e.g., you absolutely need to have computationally intense stuff like 3D graphics or machine learning).

Common installation packages, the ones that tell you to shut down everything else and make you click next 50 times just to install one little app, suck and could be improved.

I don't see how this corresponds to "the desktop is dead".

If your badass uber-leet Ajax app takes 20+ seconds to load, uses up 80-140+ MB of RAM, feels slow on a Pentium 4 machine, does less than an Apple IIe app, and could go down at any time leaving you totally helpless (as happened the other day in spite of the high-speed "backup" systems those high-profile Web 2.0 sites allegedly had), you have failed. Deal with it.

You can use simple reality checks here. If your clunky "rich text" editing widget thingie is crippled in ways Notepad is not, you have failed. If your spreadsheet app is maddeningly frustrating to use for more than 100 rows or if you know any keyboard shortcuts at all, you have failed. And so forth.

Hyperlinks are better than buttons. I think the blending of information and functionality in web apps is superior to desktop apps, and I'm also a desktop app developer. So many things that are inherently superior in web apps... the web browser platform is a very mature text and graphics rendering engine. Updates happen automatically and seamlessly. I could go on, but its not just .NET

Hmm, let me think about the primary apps people around my office use when they want to get work done (instead of goofing off on the web). Artists: Photoshop, Illustrator, etc. Coders: Visual Studio, Eclipse, Emacs, etc. Project Managers: MS Project, Requisite Pro, JIRA, etc. Business/Marketing: Outlook, Excel, PowerPoint, etc.

Only one non-desktop app in the mix there - that being JIRA. Looks like the death knell was a tad premature once again.

Good article overall (its main message that is), but there were too many generalizations that either aren't true or just stupid:

ex. "After all, people have managed to learn how not to put everything they see in their mouth, learned how to stop on red and go on green... They'll learn not to drop viruses on their desktops."

ummm way easier said than done given the stats and me watching my mom's generation use computers... not to mention how it's a lot harder building something stable and secure as opposed to trying to break it (not to mention MS's general failure at building good stuff)

"Do not point your finger at .NET, it is still a subset of Win32. It is built mostly on top of Win32, and it is still not available for us to use! Yes, .NET runtime is not part of the most popular Desktop out there - Windows XP."

This didn't happen out of stupidity, MS wanted to bundled it. What stopped it was a thing called an anti-trust lawsuit... (.NET was made to kill Java)

It would be nice if desktop apps could be write-once, run everyowhere important -- like web apps kinda are. Easy client-server communication on a desktop app for persistence between computers would also be nice.

There is something to be said for really peppy interfaces, rich input mechanisms (multi-touch, joysticks, gestures through a webcam), and compiled graphics libraries that blow our minds.

There is probably room for an open source system that uses a good languages, provides these capabilities, and could run everywhere. A model where monetization comes from support could probably work.

Something like QT is clearly not there yet. Doesn't actually run everywhere and C++ stinks.

Tcl and Tk were actually a pretty good effort in that direction, until Tcl's marketing fell apart and they took something like 8 years to release a version of Tk that didn't look bad and out of place on Unix.

Still though, there's no contest with web applications in terms of how easy it is for the developer to deploy them.

Tcl and Tk are ok. Not the best.

I think nothing is dead.We are in a phase that exploring the possibilities of huge network connectivity,distributed network services, the power of collaboration in content creating,knowledge shearing etc...

I think the third phase will be somewhere in between desktop app. - web app. When that time come, we'll only need differrent kind of browser (or browsing experience) which present power of desktop app. and backed with the information rich web app. services and techniques.(3d surfing,clear navigation of web space in 3d,"web space" as a real 3d space experience)

remember, life don't likes extreme sides.

Apparently the reason Microsoft is dead is that Microsoft is suicidal. How 'bout that.

Your OS, your web browser, your ISP's server apps etc. are all software. Software isn't dead, nor will it die too soon. It's just in a phase of transition.

each has its own pros and cons! one thing i feel is that browsers should rethink about some features like desktop access. This is will help in more powerful apps.

I'm guessing not too many of you use MSFT outlook. I could seriously load up yahoo and google mail 10 times each before outlook is running. There's a certain performance advantage to webapps...the GUI stuff (the real code that renders graphics) is always in RAM. The full process from hitting gmail to displaying email requires very little hard drive access.

I use outlook. In fact, I just measured the load-up time for Outlook and gmail.

    Outlook:  12 seconds
    Firefox+Gmail:  21  seconds.

Finding and opening messages is much easier in Outlook. I can use tab and arrow keys to navigate much easier than I can with gmail, and messages load much more quickly. Since it's an Exchange server that maintains a constant connection, I get email from anyone on the network within a few seconds, and a notification in the taskbar that I've got a new message. Mail filters are not difficult to set up, and arranging messages is a drag-and-drop affair.

I'm using Outlook less these days, as my co-workers tend to use IRC for most incidental communication and email for technical notifications and documentation and such. But at my last company most professionals, managers, and executives lived in Outlook.

There are a few things that really bug me about Outlook: No support for tabbed browsing of messages and a buggy, poorly documented scripting engine. Practically speaking, Outlook requires more IT support than gmail.

But once Outlook's up, it's got a much better user experience, so far as I'm concerned. There's a long list of features that Gmail just can't match that come from being a desktop app.

Hard drives transfer data faster than ISP's in the first place, and why have a hard drive if you're not going to use it?

but do most people need those features? most of it is just bloat made for a marketing brochure in my opinion

It's more of a long-tail effect. Outlook has a lot of features, and it's a bit bloaty in the sense that they could be better organized but having supported people using outlook for a couple of years I can say that people find the Way They Like To Do Things and it's not always the same as everyone else. People find the feature they like and use it. Delegate configurations are very common executives and their assistants. A friend of mine used the notes feature all the time.

If you actually look at how people use email you will find quite a bit of diversity. However, one of the common threads in email usage is task management. This highlights one of the biggest areas of deficiency in Gmail. Take the lack of calendar integration as one example.

Sure. Here's one: I can drag and drop an attachment for an email right onto it. No can do in gmail.