Google: The internet is 'the right programming model' Just five years ago, Vic Gundotra argued that web apps could never rival their desktop brethren. But that's when he worked for Microsoft. He works for Google now. And he sees things quite differently. In 2004, Gundotra and his Microsoft team - responsible for driving developers to Windows - pointed to an application called …
Nice new 'no' clothes.
Let's see now.....
ANDROID will become the new KERNUL. CHROME will become the new OS. HTML5 will become the new HLPL. And.... when you work OFFLINE you will be using your locally stored HTML5 progrums [Fluffy Toys].
Hmmmmmm
METUL
KERNUL
OS
HLPL
PROGRUMS
Well Fuck Me if that isn't what we haven't got at the moment. Better call it a Pardugm then.
This is the same old centralized server farm versus distributed/point to point argument.
We've seen this pendulum go back and forth so many times I've lost count.
Ultimately, assuming performance is king, it's hard to see an end game other than local "fat" apps which are updated transparently. Much like Ubuntu's package updates, this could be automatic and pain free. Hopefully an application sandbox similar to that used by Java applets will become the norm such that it should be possible to run any untrusted applications without fear of comprising system security. A new network file system wouldn't hurt either.
If only the industry could settle on portable application specifications for next generation computing, then the "run anywhere" advantages of HTML & HTTP, which is a stretch to begin with, could be overcome by local apps.
Google has shown that with enough money, a centralized model can be made to work most of the time, even if their true reasons are control over user data and lock in. However sooner or later a competitor boasting more efficient local/point to point technologies will come to market and swing the pendulum the other way again.
Google wants us to believe that cloud computing is the natural order of everything in programming only because Google's business model hinges around cloud computing. This may be true for certain applications (spreadsheets, word processing, etc.), but the reality is that the world is quickly moving toward a massively automated age in which cars can either drive themselves (or, at least, react to dangerous situations to prevent collisions) and personal robots do chores around the house including the baby sitting. The computing horsepower that will be required for this sort of stuff can only be achieved via massive parallelism and gargantuan random access memory. There is no way the public is going to trust their mission and safety-critical computing needs to an invisible cloud in the sky. What if the system suddenly becomes overwhelmed with processing requests or some technical difficulties? Imagine the damage that a cloud blackout or brownout could do to a modern economy that is completely hooked to it.
So no, Eric. Cloud computing is a very good thing but it is not a programming model. It's just a business model. As soon as some maverick startup comes out with a solution to the parallel programming crisis (see link below), get ready for an explosion of super-complex and environmentally-aware apps that would tax the bandwidth of any cloud infrastuture.
How to Solve the Parallel Programming Crisis:
http://rebelscience.blogspot.com/2008/07/how-to-solve-parallel-programming.html
Man, who used to work for hard drive based software company and used to proclaim same was "the way" has moved job to a web based company and now claims that is the best way.
And this is news how?
Google are not to be trusted.
Must you wake up in a world where they control everything BEFORE you realise?
Yes I use some of this services, some of them more some of them less.
But at the moment I can't imagine to put all my data in the net.
Ok e.g. e-mails on Gmail is already at the edge but my finace data or contracts, my health information or other private things
I don't believe at the moment I will store this stuff in future somewhere in the cloud...
And wan't it first used for Exchange web access.... which pretty much points to MS beating GOG to the punch by a few years in terms of a async xml javascript mail goodness by some 4 years? Not that I think it was a patch on Gmail, but do think that MS gets the web in many ways like GM gets electric cars. They have been leaders, have good ideas for the future, but its about 1% of their revenue stream, so lets not rock the boat shall we.
""It works no matter what device you're using, no matter what operating system you're using, whatever operating system you're using - as long as you're connected.""
Wow! How totally cool! Google's invented TCP/IP! Sign me up![1]
"But then he caught himself. "Even then, if you're offline, it works with some caching.""
Oh. Darn. It's just repackaged UUCP. Do I have to download & compile Kermit again?
[1]Pardon me while I shower after channeling Yahoo! ...
...same as the old king.
Every month, I see more and more data handed over to remote third parties through the medium of web apps. And every month, I see the client side of these web-apps become larger and heavier to compensate. You only have to look at conceptually borked models such as AJAX and all the increasingly passionate talk about Javascript and its flexiblility to see this.
We're going to end up with "web apps" that are as heavy on the client side as old-fashioned local programs with the sole difference being that they store their data off-site. The only remaining differences are whether you have to install the software to use it and whether you can access your data when you're not connected.
Oh wait, we are now starting to see web apps that you have to install, and with caching, we are now seeing our data kept locally. It seems to me that all that will be achieved is the sharing of your personal data with third party companies you have no control over.
Paris - because at least she knows when she's been fucked, unlike all the Google enthusiasts.
I rather suspect this has something to do with the way that every single browser seems to interpret "standards" differently. Using javascript or CSS to do anything complex has always been a nightmare... you could either go for the broadest market (ie, IE) and have a lousy platform (because activeX is better, naturally), support a less widely used browser (or browsers) and alienate many potential viewers, or do something else like use a vast, bloated and slow javascript library. Or avoid the whole thing altogether and use flash.
HTML5 isn't going to fix the fact that programming for the web is a particularly painful task, and probably always will be.
That's the message I got out of the article, anyway.
It all sounded just like Microsoft some years back! "We're huge, we're great, we've got the best product, we've got the best strategy and anyone who doesn't agree with us is dead in the water."
Remember here, Microsoft started with the attitude that because Windows was so bloody superior to everything else, anything not using it would die. They eventually had to resort to good, old-fashioned dirty tactics when this proved untrue.
It's not just the thin/fat client thing that repeats itself.......
...Google decide they no longer want to support the app and close it down, or they do an upgrade that makes it incompatible with your systems.
Never happen, look how many services Yahoo, Lycos and Microsoft have decided to close down becuase they no longer fit their business models.
Do you think they will listen to you?
Thought not.
My Hardware, My Software. I choose when and what to upgrade (if ever).
That seems to be the main problem to me - just as the move to microcomputers involved jettisoning managed memory, multi-tasking, and high-level programming languages, so this move is equally throwing out decades of desktop development.
So far, most of these web apps are silos - they are like desktop apps in the bad old days before universal cut and paste, or if you're on a decent OS, things like user-extensible system-wide services. (Sorry, I forgot - we have Greasemonkey for that).
Solving that problem is not going to be easy (security, trust between applications, etc).
Then there is the whole question of programming languages and frameworks - the browser seems roughly equivalent to something like a graphics layer, but not a lot more, hence the emergence frameworks on top of JavaScript to offer basic 'application' structure, etc . . . but as we know most of these involve downloading a few hundred K of JavaScript (or Flex libraries).
Again, this feels like a huge step backwards (every app needs to include all the framework code). Not insoluble (see Adobe's RSL approach for saving signed shared Flash libraries) but it says a lot about the gulf between the browser APIs and any modern app development framework.
But, you know, all that stuff is just old and boring.
Sure it's the next big thing.....yeah, maybe when we can get more than 512k to our offices - PAH!
I hope Google isn't investing too much of it's good empire in this stuff because I honestly don't think it will catch on, I don't think the masses will buy into it. I know I'm looking at it with sceptical eyes simply because I like my critical data and systems local, where I can keep an eye on them.
Perhaps that will become an outdated and old-school view, but I wouldn't care if it did.
Also, this Vic Gundotra guy is plainly just doing his job and generating a bit of hype, of course his opinion can't be trusted. He's simply shouting whatever his current employers expect him to shout.
"Google wants us to believe that cloud computing is the natural order of everything in programming only because Google's business model hinges around cloud computing"
And Savain wants us to believe that local computing is the natural order of everything in programming only because Savain's business model hinges around local computing.
Savain, presumably the reason you don't like the cloud model is because your model of "parallelism" doesn't support distributed computing (to directly quote one of the more ridiculous statements on your site, "In a truly parallel system, all events are synchronized to a global clock").
Meanwhile, despite the "parallel programming crisis" you keep insisting on, Google are writing huge distributed parallel applications that your model cannot even approximate.
"http://rebelscience.blogspot.com/2008/07/how-to-solve-parallel-programming.html"
The irony that your blog is hosted in the cloud by Google is not lost on me either.
Wow, I see lots of comments deriding Google's plans and their same-ol'-same-ol' view of the centralized vs. point-to-point paradigm. That's great! I guess I'm not the only one who distrusts Google.
However, they only seem to see part of the picture. This is just Google putting up a puppet show to distract everyone from their real plans. While everybody's enthralled with the puppets and their shiny toys, they hope not many are paying attention to their data retention policies, their government lobbying for a sanction monopoly on web data access, their land-grab of available information, and more importantly, their consolidation and centralised control of advertising technologies.
This is Microsoft's Hailstorm all over again; even Overture's data-collection dream! Google wants to control access to information--all information: health care records, books, tv, music, historical and geographical data, consumer purchases and browsing habits, etc.--and put ads in them. That's where the real money is at! They don't want to *own* the data; they want to *control* access to it, and own the right to sell advertising for this.
Where those other companies failed was in grabbing everything too fast, attracting public and government attention, and ultimately scorn. They did not have a puppet show to distract the masses. They were powerful corporations attempting to gain ultimate control on data access and sharing, plainly as day. We all saw it, and the governments of the world saw the money train and wanted their piece of the pie as well.
The real danger here is to keep viewing Google as the underdog, the innocent new comer, the small and agile software innovator. It is none of these. They are giving away access to their web applications because the real money is in the advertising business, and as long as they keep pumping out free software people will keep thinking they are just a software company. An innocent software company who can Do No Evil (tm).
But don't you pay attention to that, here's a little puppet with a shiny toy... for free! Look at the puppet...
-dZ.
If an application works when it is cached on the local machine that ... errr.... makes it a locally running program.
The web will never win the war when it comes to applications. People won't let it, because the cost model threatens to run away as opposed to a known, fixed cost for a package which can be used as long as the operating system isn't upgraded to an incompatible version.
Come off it - javascript was pulling dynamic content from servers years before someone decided to slap "XML" into the name of the technique to sell more powerpoint presentations. What made the difference was the DOM settling down so that code to do these things could be relied on. XMLHttpRequest is not needed and never was - it's just a marketing buzzword. Rather like XML itself, really.
This post has been deleted by its author
Louis Savain 'business model' comprises the demand that someone (intel is oft mentioned) give him some millions of dollars and a couple of years and he'll deliver something amazing.
Shame he can't even get anything working until then, innit.
Louis, you're not "some maverick" but a monkey. Some of them can even make their own tools (<http://www.newscientist.com/article/mg20227104.300-sweet-tooth-drives-tool-use-in-chimpanzees.html>). Apparently you can't.
The Register 
Biting the hand that feeds IT
Copyright. All rights reserved © 1998–2024
