It seems like the market for browser plugins is thriving. Firefox, Chrome, IE, and now even Safari all support them, and some are popular enough that even non techies are likely to have them.
In general, plugins are a cool and useful capability to have around, but I think they are also holding the web back.
Think about flash. A lot of ink has been spilled lately about the flash versus html5 issue. But had browser plugins never been a possibility, we would have had to develop a rich, standards compliant canvas tag long ago, because the only vector for innovation would be better html features.
I fear the same thing will happen to browser-side databases. With the likelihood diminishing that HTML5 will be able to specify a database standard that pleases the required parties, we might be left to pursue the “make up” option of the (now defunct) Gears-style database — all the goodness of a client-side database, but wrapped in a browser plugin. And just functional enough to keep us from feeling the urgency to standardize the technology.
I say throw away these system-level browser plugins and lets return to the days of no-holds-barred unauthorized HTML extension. Let Chrome and Firefox support incompatible supersets of HTML, because only then will experimentation be done in a way that is both disrupting — the losers will have no choice but to implement the standards of the victors — but also open. Because when you compete by extending HTML (instead of writing a plugin), you do so in a way that anyone is able to reproduce if they decide it is worth it to them. And that is far better than the closed world of browser plugins.
History is filled with alternating cycles of stability and disruption. After a good run by the Church of Web Standards Orthodoxy, I think we’re ready for some disruption.