There are many considerations I believe you have left out, and this leads to an overly fast dismissal of the language. The decision is much more nuanced than calling it a "stumbling block" stemming from Many Eyes' IBM heritage.

First, remember to think from the developer's perspective. Most everyone, both academically and professionally, knows Java to some extent. Note how a few researchers banded together and made a product with so many superior characteristics to the fancy startup's version. Java is significantly more mature than AJAX or Flash. Rapid, quality development certainly has its advantages.

Second, think about long-term product integration. In what ways can Swivel capitalize on its sunk costs and investment in its codebase? AJAX is mostly bound to the internet, though there are ways to use it locally. Instead, it makes much more business sense to create a solid Java product that could be integrated into any environment, from web to enterprise desktop.

Third, the maligning the Java install base makes little sense in comparison to competing technologies. JavaScript has quite a few problems of its own, especially in compatibility across browsers. Yes, all browsers have some amount of it-- but programming to its moving target is a whole can of worms in itself (see #1). Thus, its installed base cannot be considered as pervasive as first glance would imply. Flash is even worse; my experience gives it very low performance and many technical difficulties, making it still much more of a niche product. AJAX is not the Internet's silver bullet.

I believe the fundamental fallacy here is lack of context. Swivel wants to be on the Internet exclusively; IBM would prefer not to have Many Eyes tied down to any one medium or take too much effort to flesh out quickly. They want maximum returns on research investment, making Java a no-brainer. It is only from our personal, limited perspective that the technology appears sub-optimal.

That said, I complement you on an excellently written article.