A million is probably still small compared to what they would make off a revolutionary algorithm (and significantly smaller than MS would make off a better spellcheck), but at least it's enough to quit your job and do what you want for a while.

10k is nothing. It's an insult to the work that would go into an improvement.

If you've already got a better algorithm, you now know it's worth at least $10,000 to Microsoft.

Debatably. This could be a talent-scouting endeavor, so the person who came up with it (and thus this and their future creations) could be worth at least 10k to MS.

Though I think it's more likely a side-goal, so I do agree. Just pointing out alternatives :)