Thanks a lot for the detailed reply. I think the best strategy for me is option (c).
I agree that slide decks suck as a sharing medium (unless you go to the trouble of adding "notes" under each slide, and share that, but then you're better off sharing a different document). Martin Fowler puts it very well in this post: http://martinfowler.com/bliki/Slideument.html
I can see how sharing a slide deck would dilute the value of a talk. At the same time, sharing my slides lets me reach more people. They might learn more by watching the talk, but it's a trade-off.
There's also the question of the legality of sharing these slides on websites such as InfoQ or Parleys, where the slides are synchronized with the video. But I guess option (c) could work for these.
One thing to note is that converting to outlines will essentially change the text to an 'image'. So it's like looking at a jpeg of text - it won't be able to be highlighted, copied etc. (I'm sure there was a better way to word that, but the terminology escapes me)
Given how it's a slide deck, this probably isn't a big deal for distribution. But if it's important to you, I'm sure you can OCR the final product, and that wouldn't end up embedding the fonts. Unfortunately that's sort of a roundabout process.
It's true that you can't treat the type as text anymore (if you can select it, it's probably not OK to share the file), but, importantly, you can scale it to any size with perfect fidelity. That's what you don't get in any process involving round-tripping through a JPG.
This is surely an annoying process bound to put anyone off of commercial typography... unless you're already typesetting your presentations in Illustrator. Which, admittedly, I haven't been doing. But now that I think about it: I do a lot of stuff in Illustrator, and Illustrator is so. much. better. than Keynote.
Good point about not being able to copy/paste the final text.
I don't understand why I should OCR the final output, though, since I already have access to the "raw" text. If I were to OCR it and generate a new PDF, I'm back to square one, since I'd need to specify a font. Unless you meant that some people might want to OCR my public PDF to access the underlying text, since it only contains images?
In the end, I guess I'll simply use a font with a more permissive licence (free or commercial).