3 ms·

Of course you're right that some domains (e.g. graphics) are harder than others, but you might be overstating your case. Surely Gmail requires "non-trivial inte

by gruseom 14y ago

Of course you're right that some domains (e.g. graphics) are harder than others, but you might be overstating your case. Surely Gmail requires "non-trivial interactions". And why not mathematical modelling/simulation? — you can access a lot more computational power on servers. And how often does one hear of a startup making new desktop software? I like your point that we might figure out how to get desktop deployment right sooner than how to adapt the web to do all these things. But nearly all the innovation has been pointing in the other direction. Also, such a deployment technology would face formidable institutional barriers.

timburks 14y ago
The Gmail native iOS app is really nice, so the flow of innovation isn't consistently pointing towards the web. But part of the answer to getting "desktop deployment right" is sandboxing. In essence, the native app platforms are making operating systems more like web browsers, while projects like Mozilla and Chrome are trying to turn web browsers into operating systems.
Silhouette 14y ago
Surely Gmail requires "non-trivial interactions". Yes and no. GMail is arguably one of the most successful and mature web apps out there, but at the end of the day you're still just reading mail, clicking a few places to change how information is presented, and typing a bunch of text. None of this requires a lot of real-time assistance as you navigate the information, context-sensitive UI, or task-specific controls beyond regular menu/toolbar/icon kinds of things, for example. And why not mathematical modelling/simulation? — you can access a lot more computational power on servers. Well, I think that point is debatable if you're doing serious modelling more than occasionally. The kind of processing power required makes it cheaper to buy your own high-end systems than to use commodity cloud-provided facilities very quickly, and for the same reason I suspect it would be difficult to offer such services economically in SaaS form. Also, real time graphic visualisations in web apps are still far slower and less sophisticated than what you can achieve natively with modern graphics cards, at least for now. If you're watching a couple of pretty graphics in an analytics tool, that doesn't matter. Even some basic 3D work is starting to get interesting with technologies like WebGL arriving. But for real-time rendering and navigation of global weather patterns or the design of a cruise ship? I don't see that being done remotely any time soon. And how often does one hear of a startup making new desktop software? I'm not sure how often one hears about start-ups doing something is a very useful metric, really. The culture on HN is particularly biased towards that field, but in reality there are millions of programmers out there working on desktop software. New bespoke desktop applications are being written for businesses all the time, and likewise tools, toys, games and more are written by small software shops for individuals all the time. But nearly all the innovation has been pointing in the other direction. Really? I'd say the mobile "app store" model and the Linux "distribution" model have both been very successful at getting native software deployed onto users' systems, too. It's just a shame that we still have horrible deployment practices and dependency management on most desktop platforms, much of it due to the stagnation of desktop OS architectures and the continued use of strategies that were originally designed to cope with limited resources but that have been obsolete for years now.
- gruseom 14y ago
  I think you underestimate how much is going on in an app like Gmail, and also the extent to which people have figured out how to do things like "real-time assistance as you navigate the information, context-sensitive UI, [and] task-specific controls" in the browser. Those problems are mostly solved. Some things, like rich text editing, still suck. I'm not sure how often one hears about start-ups doing something is a very useful metric, really. Startups aren't what drives tech innovation in markets? That seems at odds with history. Millions of programmers may be doing as you say, but that's mostly legacy. Are they creating fundamentally new things? What is the most recent desktop app that you could call innovative, apart from pushing the envelope on graphics? I'd say the mobile "app store" model and the Linux "distribution" model have both been very successful at getting native software deployed onto users' systems, too. I guess Apple have integrated desktop deployment into their App Store (rather to my annoyance actually) and MS are about to do the same, or try to. Perhaps the habits people have acquired around downloading and installing mobile apps will end by triggering a renaissance in desktop applications. But I won't believe that I until I see major new applications being built this way.
  - Silhouette 14y ago
    I think you underestimate how much is going on in an app like Gmail I've no doubt they do all kinds of clever things behind the scenes, from detecting things like malware and spam to all the analysis Google do for their own purposes. Still, relatively speaking, I don't think the front end for an e-mail client is very complicated. And there's nothing wrong with that, because it doesn't need to be: its job is to manipulate a small amount of data (from the user's point of view, with a simple structure, which doesn't change very much and doesn't have a lot of dependencies. This is all true of most major web apps, as far as I can see. ..."real-time assistance as you navigate the information, context-sensitive UI, [and] task-specific controls" in the browser. Those problems are mostly solved. Don't you think "mostly solved" is overstating the case a little? Even web apps whose UIs we might call mature or advanced today rarely offer more than what desktop WIMP UIs were doing 10 or even 20 years ago. As you mentioned, even common operations like editing formatted text are often still below par. Truly task-specific UIs, the equivalent of say a three-way diff/merge tool or a space flight simulator game on the desktop, seem to be extremely rare in web apps. Startups aren't what drives tech innovation in markets? Start-ups can certainly drive innovation, though obviously this particular forum has a built-in bias and I think a lot of posters here tend to overrate the importance of any one start-up or of start-ups generally. However, I don't think whether or not we've heard of start-ups means very much. It's a big world, and most of us will only ever see a tiny fraction of it. Lots of the innovation happens in small businesses that help larger businesses or other non-consumer fields like academic research or government departments behind the scenes. These things don't tend to be very visible to the general public, and probably won't get you on the front page of HN or written up on TechCrunch. Plenty of these projects involve developing desktop apps. What is the most recent desktop app that you could call innovative, apart from pushing the envelope on graphics? I can't give personal examples from my own work unfortunately, so instead let me ask you this: have you been to a movie lately? Did you notice all the CGI special effects? Not the robots and zombies and futuristic home automation, the other ones. The answer is probably "not even close", because the programs the SFX wizards use to create those effects are now so powerful that the results really do go unnoticed a lot of the time, as they should. These things are done with a vast array of bespoke software that is evolving all the time, some of it off-the-shelf but a significant chunk built in-house for major projects. And they're not building web apps. (Most of the time. ;-)) Perhaps the habits people have acquired around downloading and installing mobile apps will end by triggering a renaissance in desktop applications. But I won't believe that I until I see major new applications being built this way. I tend to agree. I fear that ship has now sailed, but it would be nice to think that the groups behind the significant desktop platforms could still come up with a sane way to install and update applications. It's crazy that in 2012 (just...) we still rely on each application's installer/uninstaller/package manager/update process to copy files into basically arbitrary locations, some of which may be shared with other software, set up hidden details like registry entries, even hack around global system configuration or perform dark magic for product activation/DRM purposes. And then we expect to install more than two things and have them play nicely together, and then uninstall in some arbitrary order and still have them play nice together. Here's hoping 2013 will bring something better!

by gruseom 14y ago

timburks 14y ago

The Gmail native iOS app is really nice, so the flow of innovation isn't consistently pointing towards the web. But part of the answer to getting "desktop deployment right" is sandboxing. In essence, the native app platforms are making operating systems more like web browsers, while projects like Mozilla and Chrome are trying to turn web browsers into operating systems.

Silhouette 14y ago

Surely Gmail requires "non-trivial interactions". Yes and no. GMail is arguably one of the most successful and mature web apps out there, but at the end of the day you're still just reading mail, clicking a few places to change how information is presented, and typing a bunch of text. None of this requires a lot of real-time assistance as you navigate the information, context-sensitive UI, or task-specific controls beyond regular menu/toolbar/icon kinds of things, for example. And why not mathematical modelling/simulation? — you can access a lot more computational power on servers. Well, I think that point is debatable if you're doing serious modelling more than occasionally. The kind of processing power required makes it cheaper to buy your own high-end systems than to use commodity cloud-provided facilities very quickly, and for the same reason I suspect it would be difficult to offer such services economically in SaaS form. Also, real time graphic visualisations in web apps are still far slower and less sophisticated than what you can achieve natively with modern graphics cards, at least for now. If you're watching a couple of pretty graphics in an analytics tool, that doesn't matter. Even some basic 3D work is starting to get interesting with technologies like WebGL arriving. But for real-time rendering and navigation of global weather patterns or the design of a cruise ship? I don't see that being done remotely any time soon. And how often does one hear of a startup making new desktop software? I'm not sure how often one hears about start-ups doing something is a very useful metric, really. The culture on HN is particularly biased towards that field, but in reality there are millions of programmers out there working on desktop software. New bespoke desktop applications are being written for businesses all the time, and likewise tools, toys, games and more are written by small software shops for individuals all the time. But nearly all the innovation has been pointing in the other direction. Really? I'd say the mobile "app store" model and the Linux "distribution" model have both been very successful at getting native software deployed onto users' systems, too. It's just a shame that we still have horrible deployment practices and dependency management on most desktop platforms, much of it due to the stagnation of desktop OS architectures and the continued use of strategies that were originally designed to cope with limited resources but that have been obsolete for years now.

gruseom 14y ago
I think you underestimate how much is going on in an app like Gmail, and also the extent to which people have figured out how to do things like "real-time assistance as you navigate the information, context-sensitive UI, [and] task-specific controls" in the browser. Those problems are mostly solved. Some things, like rich text editing, still suck. I'm not sure how often one hears about start-ups doing something is a very useful metric, really. Startups aren't what drives tech innovation in markets? That seems at odds with history. Millions of programmers may be doing as you say, but that's mostly legacy. Are they creating fundamentally new things? What is the most recent desktop app that you could call innovative, apart from pushing the envelope on graphics? I'd say the mobile "app store" model and the Linux "distribution" model have both been very successful at getting native software deployed onto users' systems, too. I guess Apple have integrated desktop deployment into their App Store (rather to my annoyance actually) and MS are about to do the same, or try to. Perhaps the habits people have acquired around downloading and installing mobile apps will end by triggering a renaissance in desktop applications. But I won't believe that I until I see major new applications being built this way.
- Silhouette 14y ago
  I think you underestimate how much is going on in an app like Gmail I've no doubt they do all kinds of clever things behind the scenes, from detecting things like malware and spam to all the analysis Google do for their own purposes. Still, relatively speaking, I don't think the front end for an e-mail client is very complicated. And there's nothing wrong with that, because it doesn't need to be: its job is to manipulate a small amount of data (from the user's point of view, with a simple structure, which doesn't change very much and doesn't have a lot of dependencies. This is all true of most major web apps, as far as I can see. ..."real-time assistance as you navigate the information, context-sensitive UI, [and] task-specific controls" in the browser. Those problems are mostly solved. Don't you think "mostly solved" is overstating the case a little? Even web apps whose UIs we might call mature or advanced today rarely offer more than what desktop WIMP UIs were doing 10 or even 20 years ago. As you mentioned, even common operations like editing formatted text are often still below par. Truly task-specific UIs, the equivalent of say a three-way diff/merge tool or a space flight simulator game on the desktop, seem to be extremely rare in web apps. Startups aren't what drives tech innovation in markets? Start-ups can certainly drive innovation, though obviously this particular forum has a built-in bias and I think a lot of posters here tend to overrate the importance of any one start-up or of start-ups generally. However, I don't think whether or not we've heard of start-ups means very much. It's a big world, and most of us will only ever see a tiny fraction of it. Lots of the innovation happens in small businesses that help larger businesses or other non-consumer fields like academic research or government departments behind the scenes. These things don't tend to be very visible to the general public, and probably won't get you on the front page of HN or written up on TechCrunch. Plenty of these projects involve developing desktop apps. What is the most recent desktop app that you could call innovative, apart from pushing the envelope on graphics? I can't give personal examples from my own work unfortunately, so instead let me ask you this: have you been to a movie lately? Did you notice all the CGI special effects? Not the robots and zombies and futuristic home automation, the other ones. The answer is probably "not even close", because the programs the SFX wizards use to create those effects are now so powerful that the results really do go unnoticed a lot of the time, as they should. These things are done with a vast array of bespoke software that is evolving all the time, some of it off-the-shelf but a significant chunk built in-house for major projects. And they're not building web apps. (Most of the time. ;-)) Perhaps the habits people have acquired around downloading and installing mobile apps will end by triggering a renaissance in desktop applications. But I won't believe that I until I see major new applications being built this way. I tend to agree. I fear that ship has now sailed, but it would be nice to think that the groups behind the significant desktop platforms could still come up with a sane way to install and update applications. It's crazy that in 2012 (just...) we still rely on each application's installer/uninstaller/package manager/update process to copy files into basically arbitrary locations, some of which may be shared with other software, set up hidden details like registry entries, even hack around global system configuration or perform dark magic for product activation/DRM purposes. And then we expect to install more than two things and have them play nicely together, and then uninstall in some arbitrary order and still have them play nice together. Here's hoping 2013 will bring something better!