Can Web Apps Ever Truly Replace Desktop Apps?

tooger writes "Matt Hartley from MadPenguin.org opines that web apps can never replace desktop applications, for a variety of reasons. He writes, 'Some of you may point out that the data stored on your hard drive is not of any real consequence, but I would disagree. It is more than probable that a skilled, disgruntled employee of the company you trust with your data could ... sell off your personal information.' Given the real danger of privacy concerns, identity theft, and uptime, will web-based applications ever truly replace locally hosted software?"

Comments

a more appropriate question:

[yagu]

A more appropriate question might be, "What is the extent to which Web apps will be effective, and accepted?".

Many desktop applications are of that ilk solely from the era of their birth. There isn't always a compelling reason an application needs to run on a desktop, and Web offers another and slightly different alternative. And as for some of Web apps shortcomings pointed out by the author, they're mostly nits, things that will be solved soon, or already solved.

I for one find Google applications (spreadsheet, word processor) perfectly good replacements for my more modest needs day to day. They come close, at this very immature stage in their life cycles, to being able to completely replace my need for desktop instantiations. I would guess the average lay-person would fall more neatly into this demographic -- the average computer user could save lots of dollars by getting comfortable with the scaled back versions of stuff they paid big money for but never tapped the deep and myriad powers from.

There probably always will be a place and reason for desktop applications: data security, data privacy, contracts, speed, availability, etc., but Web offers another approach and an increasingly viable approach to replacing applications we all once thought of as "desktop".

As a developer, it's changed my way of thinking when it comes to creating and designing new products. It isn't a hard transition, and it offers some interesting new ways to make magic for my clients (mashups, etc.).

The article describes "lack of sync" options with Google apps. Yawn. I've written my own for now, I agree it's a bit of a nuisance. Does anybody think for a moment these gaps aren't going to be filled soon?

Re:a more appropriate question:

[MightyMartian]

The major advantage of web apps is distribution. With browsers adhering more closely to the standards, the difficulties involved in making cross-browser apps is less difficult, and makes distributing such apps far cheaper and easier than desktop apps.

That being said, I think we're a long way off from having browser apps that can really compete with their desktop equivalents. Even a highly usable site like GMail still is awkward and clumsy compared to your average mail client. Google Docs is interesting, and for simple work does a reasonably good job, but still doesn't have the responsiveness of using Word or OpenOffice.

Security is a big issue as well. You have to take the word of the company that your data is being held safe and secure. I simply wouldn't trust Google to keep my data secure at this point. Though Windows has no lack of security problems, at least the computer is in front of me, and there are number of ways I can assure that I'm protected. Google is a black box. What happens beyond the server is beyond my capacity to tell, and I have no way of assuring that it's safe and secure.

Re:a more appropriate question:

[nine-times]

The article describes "lack of sync" options with Google apps. Yawn. I've written my own for now, I agree it's a bit of a nuisance. Does anybody think for a moment these gaps aren't going to be filled soon?

If you've written your own, maybe you want to open the source code to everyone else? These gaps might be filled soon, but they aren't filled yet. Besides, the problem isn't just "syncing", but rather a lack of consistency.

If I'm working on a Word document, I can upload that to Google, and I'll probably lose some formatting and features. Then, once on Google's apps, I have to work in a different interface. Then I have to make sure things are syncing constantly, and finally I can export it back into a Word document and I'm back working in Word with a different interface and feature-set again. Plus, I have to continually sync everything to make sure my copies aren't out of date. And then, finally, what happens if I accidently edit both the local and remote copy? Where is my tool for reconciling these?

What really needs to happen is some sort of melding between web apps and local apps. If I'm going to use Google's word processor, I need to be able to cache everything locally, including the documents and the application itself, so that I can continue to work when my internet connection drops. If I go on a day trip far away from the Internet, I need to still be able to edit those documents locally without planning ahead. Finally, if both the online and local copies get edited, Google needs a tool that can display the changes and let me reconcile them.

What's not clear to me, however, is whether a traditional web browser is the appropriate solution for this. It may be that, rather than making "web applications", we need a different framework that allows the sort of flexibility that web apps allow but with the consistency of desktop apps. Maybe a web browser should go back to reading static HTML and a new sort of generic remote application framework needs to be developed-- but who the hell is going to do that?

Re:a more appropriate question:

[dreamchaser]

There isn't always a compelling reason an application needs to run on a desktop

There are even fewer compelling reasons to run most applications via the web though. Frankly, I think web based apps are more at home on the Intranet than the Internet. The data security will *never* be quite good enough for me to trust any even remotely sensitive data to a Data Center not under the control of my organization, be it family or corporation.

Re:a more appropriate question:

[deuterium]

What's not clear to me, however, is whether a traditional web browser is the appropriate solution for this. It may be that, rather than making "web applications", we need a different framework that allows the sort of flexibility that web apps allow but with the consistency of desktop apps. Maybe a web browser should go back to reading static HTML and a new sort of generic remote application framework needs to be developed-- but who the hell is going to do that?

I'm with you here. The browser does certain things well, but has many failings as an application host. Some of my major gripes:

• No global memory scope. One of the primary tasks of an application is to store state. Currently, this is faked via cookies or server sessions, but having it so inaccessible to the client script makes it hard to emulate an application.
• Poor multimedia support. If you want to display and manipulate things like video or PDFs within a browser UI, you're wading out into no man's land of plugins that may or may not work the way you want.
• Javascript. I'm warming to it a bit, and you can do cool things with it when you know what you're doing, but it's just too abstract and incomplete to be a basis for application libraries. Scant typing, primitive array features, few math/time functions, slow.
• Reliance on a network and server. If something goes wrong with either, or if you don't have net access, you're up a creek.
• Tight sandboxing. This is a good thing, for the most part, but there are times when it makes sense to access local resources.
• Limited UI. No chance for things like 3D apps. If it can't be done with text and images, you need a plugin.
• Incompatibility. I can't count the number of hours I've spent trying to figure out why a div is 3 pixels off in IE or doesn't flow right in Firefox, etc.

Those are just some of the problems with the web as the future of apps. Web apps make sense for things involving database access and forms, but are really stretching when it comes to doing something responsive and complex, like a Photoshop or CAD program.

Traditional applications will be around as long as people want to do things a browser can't (or does poorly). It's amusing to me that as PCs have gotten ever more powerful, we're using that power less and less.

Re:a more appropriate question:

[MBGMorden]

I think you've hit the nail on the head with distribution.

As a user, and in my personal life, I HATE web based apps. I avoid them like the plague. They take my data out of my hands, often have an advertisement thrown in somewhere, require an active net connection at all times, and first and foremost they simply don't feel as responsive as a desktop application. There's also a lack of consistency. For example, for my online banking I'm pretty much forced to use the web apps that the banks use (no desktop equivalent available). I have accounts with 4 different banks. All of them have basically the same functions, but I have to learn 4 different web apps to use them. If they had a standard protocol that could tie back to a desktop app it would alleviate that problem.

On the other hand, working in IT, I love deploying and managing them. There are no software installs to perform and keep updated on lots of desktops. There are no worries about users storing important info on their local machines (even if told to store on their desktops). Also, with the progression towards these I make switching our organization over to a non-Microsoft OS on the desktop more and more possible. The more stuff that runs in the browser the less I have to worry about which OS is on the computers. And truth be told, when I'm work work, I don't really mind the problems mentioned above, because it's WORK. I don't expect my programs and computer at work to be as laid back and streamlined/comfortable as what I want at home, in the same way that I don't mind sitting in an office chair all day long but when I go home I'd toss the thing out the nearest window and get a recliner :).

That being said, the issue of data security is still an extreme concern: even when our users use a web/browser based app, it still better be running off of one of OUR servers in on OUR site.

I think that such apps will increase greatly in the corporate/government world, but that home applications will still be largely desktop in nature.

Rich Clients

[AKAImBatman]

People keep forgetting that in a corporate setting, you'll want to run your own Web Services service. While GMail for companies may make a lot of sense for the little guys, the big guys are only going to do it if they can control it internally. That takes the privacy and security concerns down to almost zero.

Re:One word answer...

[P3NIS_CLEAVER]

The correct answer is 'web apps will replace desktop apps where appropriate'

Alternative..

[zyl0x]

What if the same web-based-application architecture was applied to a corporation's intranet? This way you could host all your employees' applications in one location, as well as handle backup operations more easily.

I feel as though these "web" based applications have more than just Internet usage.

Re:One word answer...

[slazzy]

I think the best use of web applications will be to complement, not replace, desktop applications. I prefer to use a fatapp for my email because of faster searching and offline reading and writing emails - but I use webmail when I don't have access to my laptop. I would like to have the same ability with office applications.

Re:Doesn't seem to matter where it's stored

[Lockejaw]

Never mind security. If it's stored locally, I can always get at it and do what I want with it, even if I'm away from my desk and my WLAN. When I'm not at home, I can only hope that there's accessible wifi (and not one of those subscription-based hotspots).

In my opinion

[catmandi]

Personally I would identify data security as the major problem with web applications. Features, speed and availability will all improve as bandwidth becomes less of an issue - and privacy is something that could arguably be easier to control in what is essentially a thin client application. However, integrity and the (current) lack of guarantees regarding backups and recovery are the real stumbling block. If this can be overcome (and it's only going to happen when people are willing to pay for these services) then I don't see why web applications shouldn't become as popular as desktop environments.

no

[lashi]

no, just like TV never truly replace radio. They will just diverge and serve different functions in the long run.

foolish proposition anyway

[Lazerf4rt]

We are seeing more and more articles appearing with the claim that everything we really need from an OS is available online.

That's because those articles are online, and the people who write online articles love being online.

You ask the guy without an Internet connection, or with a 56K whether he thinks web apps will replace desktop apps and he'd be all like "WTF?" Keep in mind that some huge fraction of Americans never intend to get an Internet connection. Don't just dismiss that many people as idiots, either.

And how would you like it if your C++ compiler or GIMP or Photoshop or 3D Studio Max was a web application? Has anybody thought it through? It's not even a matter of security, just plain utility.

Re:foolish proposition anyway

[ACMENEWSLLC]

We need some qualifications here.

Almost all of my software I use inhouse on our LAN is web based. Mimesweeper is web based. WSUS 2.0 is web based, though for some reason 3.0 is not. My e-mail filtering solution is web based. My web filtering solution is web based. The list continues.

This software runs on my LAN at Gigabit speed. But it is web based. I can VPN in and use my web browser to run this software at home. While some of these solutions transition from Win32 to web based are great, others use Java and are thus very slow compared to the Win32 version. However, the ability to access this information from home without installing a client is great.

TS, Tarantella, Citrix allows you to take any Win32 based application and make it "web based." You can use Citrix's Internet license and put any Win32 program as a Web program. You can host this inside various Web thin clients, such as the ActiveX/Java client.

So I think moving to clients from Win32 to Internet is a good thing, as long as it is done right. Don't use Java. Google Maps is an excellent example of how well an interface can be written. Just because it's web based, doesn't mean I have to use it over a slow Internet link. It can all be loopback.

There are really two questions ...

[richg74]

Given the real danger of privacy concerns, identity theft, and uptime, will web-based applications ever truly replace locally hosted software?

There are two different issues being conflated here:

• Can an application system, in which most of the processing and data storage is done on a central host, using a Web browser as the user interface, replace conventional desktop PC applications?
• If so, will organizations and users be willing to have the hosting done by an external provider?

I would say that the answer to the first question is very probably "yes". After all, people used mainframe applications successfully for many years ; some still do. We have routinely run workstation networks with "dataless" clients (think a Unix/Linux box with only the OS, X, and swap on the local disk) precisely because we could control security and reliability more effectively. (Possibly, some users will bitch, because they want to control "their" data. If the data, as it usually does, belongs to the firm, I will punch their sympathy ticket, but otherwise -- tough.)

On the other hand, I would be wary of entrusting all my data storage and/or processing to an external provider. That raises all the same sorts of questions that any outsourcing deal does.

History keeps failing to repeat itself

[gordgekko]

Reading this discussion reminds me of every discussion I've ever seen about the thin/dumb clients vs. PC debate. Thin/dumb clients will replace the PC, why do you need all that power? Give up control! Your data is safe somewhere else!

So how is that thin/dumb client industry working out? Sell any more machines outside of a government setting since 1997?

For the most part, people want to control their important data and no serious user/business is going entrust their data to companies which promise to "do no evil" or others that have been declared monopolies or others, etc.

What does this have to do with web-vs-desktop?

[Grashnak]

It is more than probable that a skilled, disgruntled employee of the company you trust with your data could ... sell off your personal information Given that the scenario in question calls for a skilled, disgruntled insider, its unclear to me that it matters very much where your personal info is. Wouldn't the real concern be surrounding outsiders getting access to your info via a web based application? Surely an insider can get to your stuff regardless of where you keep it.

The only way...

[Mockylock]

The only way it would be likely for this to happen, would be in an enterprise situation, rather than internet and home use situations. In large companies that use application servers that hog quite a bit of resources when running apps and profiles, web based applications via the intranet could possibly use less resources and be replicated more easily than existing schemas. This can also be propagated to other web based technologies, whether it be linux OR windows based, making it much easier for companies to make decisions on equipment in the future, due to expenses. Web based applications bring about less need for higher storage in some cases, especially at the client desktop level. This will, of course, come about as a type of "linux vs. windows" ordeal, whereas each OS is fit for different customers, depending on preference. BUT, you may see companies expanding to web based options for those who it may apply to, for future customer interest.

webapps can be more secure than desktops.

[aoteoroa]

Our small manufacturing company uses desktop applications extensively, and they are a major security problem.

Sales people have quit and brought valuable proprietary information to our competitors. Giving our competitors information we worked hard (and spent a fair amount of money) to obtain.

Webapps can be secure. Your bank trusts them.

With a webapp I can guarantee that everybody has a current version of the program, that everybody is working from the most recent price lists, people can access information anywhere in the world at anytime. And when they quit they are cut off instantly. I don't have to knock on their door asking for the company laptop.

Re:webapps can be more secure than desktops.

[Anonymous Coward]

"With a webapp I can guarantee that everybody has a current version of the program, that everybody is working from the most recent price lists, people can access information anywhere in the world at anytime."

You can do all of that with a desktop app. Desktop apps can access the net too.

Re:Photoshop Replacement

[MasterC]

Then someone showed me Snipshot. Check it out, it is pretty intersting. Although it only does very basic photo editing right now, I could see where, in the future, it could support most (all?) the features of Photoshop.

Interesting: yes. I will grant it that but that's about all I'll give it. Let me know when it can zoom and edit pixels with the speed of MS Paint (let alone Photoshop or GIMP) then we can label Snipshot a real photo editor. Until then, it's a "photo tweaker" in my book.

The problem with Snipshot is that it will never attain the performance of a desktop app is because it's instructing the server to do all the work and any visual updates require sending another image back to the client after the server has performed them. The browser does zero actual work; it's the only way it can be done within the HTML/JS confines.

It will be the same issue as with video or audio but worse because both are more bandwidth intensive.

My primary complaint about any web app is speed/performance (and I'm not a performance freak, just impatient). The operations Snipshot is performing are trivial and they take a helluva lot longer than GIMP could do them in. Gmail can be dreadfully, painfully slow and is tolerable because I want the convenience.

If my prediction/opinion matters: the end result will be a hybrid with shared data. Sometimes, I just need that raw GIMP power to get crap done. Sometimes, I might be stuck on someone else's computer and not have GIMP and the handful of functions Snipshot can do may be sufficient. The marriage of desktop and web will be when I can tote those images to either app that I need them in at the time I need it. Ditto for email. I want gmail and thunderbird to sync. I want google calendar to sync with my phone and kontact. I want picasaweb to sync with kuickshow/gwenview/ee/name-your-slideshow-desktop- app.

The endgame is proper sharing of data to the app suited for the use. (Psst, just like everything else in life!) No one paradigm will "win" for every application and problem.