[glhaynes]

"We're going to have HTML5 for the next twenty years..."

Everybody's going to get up to speed with implementation of HTML5 and then ... stop? The story of HTML5 so far has been a mix of browsers chasing/leading the standards. In other words, the standards and the browsers are evolving together rapidly. I don't see why that would stop anytime soon. Certainly all this innovation won't ground to a halt when HTML5 submits their final release to the W3C (or whatever the next "formal" step is at this point).

Will there be some instance of an installed browser today that will still be running on a machine somewhere in 20 years? Well, probably... but remember that guy a couple of years ago that still had a CP/M system running his business? He wasn't holding anybody back.

And once everybody's on modern, standards-compliant browsers (and thus there's less friction for upgrading, i.e. there's no more "all our corporate apps only run in IE6"), I expect far more people will upgrade far more often. The web is at a really, really exciting time if you ask me.

So: non-Flash web technologies can reliably handle some percentage of Flash use cases today. Certainly more than a couple of years ago, and that appears to be about to go up again a lot more as HTML5 sweeps across the web. The number of use cases that require Flash are going to continue to get smaller and smaller over time. When will Flash be utterly irrelevant to the average user? I don't know. But WAY less than 20 years from now.

[benologist]

The current standards are a decade old already, HTML5 has been 6 years in the making and will probably be another 4 or 5 years before it reaches a market penetration that makes it an generally-viable design decision rather than what it is now - a device or browser specific option.

Work probably won't even start on the HTML6 specification for another decade at least while they wrap up all the formalities for HTML5, and identify where it needs improvement, and slowly discuss and agree on those improvements. Years after that browsers will start supporting it feature by feature, and years after that enough users will have access to enough features to call it mainstream.

That process is very likely to leave us having this same discussion about whether HTML6 will kill Flash in 2030. Assuming anyone still makes or cares about any of this stuff that far in the future.

Where we are now with HTML5 the void between HTML5 and plugins (Flash is just the most significant) is smaller than it's ever been before, but the fallacy of that argument is that Flash, Silverlight, JavaFX, Unity, etc are in active development and release cycles.

They make significant progress year after year, just like desktop software, operating systems, programming languages etc, and while HTML5 almost catches up right now this is as close as it's going to get at this time.

[glhaynes]

Keep in mind that they don't keep progressing linearly forever because of the nature of the service that's attempting to be provided -- neither are far off from being able to do basically all the things that can be done on a 2D panel attached to a speaker or two: interface device input, 2D/3D animation, sound effects/music, a bit of local persistence, access to the videocamera,... what else is there? I'm not sure of each's multitouch API, if any, but, really, there's not that much more beyond those things that HAS to be implemented for the vast majority of web pages and apps to fully function. Support for accelerometers and other such devices would be nice, but is hardly important to reading the New York Times or the other 99.999% of things people do/want to do with the web. Perhaps in 2030 there'll be argument about whether the web- or proprietary-way has the best support for holographic 3D displays or 360-channel surround sound, etc., but those and many other things that the two technologies don't support today seem likely to continue to overall be niche things. In other words, things whose proper role is a plug-in, not a vital part of many pages' experience. That's fine: there's going to be a role for plug-ins for a long, long time, perhaps forever. But, the fundamentals? We're getting pretty close to having them covered.

Also, I think it's well worth taking a good look at HTML5's rate of progress over the years. A huge chunk of the HTML5 effort has been toward just standardizing the response to the tag soup that is out there. That work is essentially done and doesn't have to be repeated. Just in the last year or two have things really taken off as far as adding new functionality to browsers and example pages, and it's only been in the last couple of months that a few actual commercial pages have started to use these things. In large part, that's because the web has been stuck for most of a decade because of IE stagnation. [Also note that the rest of the browser/platform vendors didn't seem to have a good mechanism to rapidly innovate together on until the HTML5 effort came together. I don't expect that that diplomatic work will have to be done again, either.] The web platform after the IE stagnation seems likely to move much faster toward covering the gaps that remain in the web platform.