Thoughts on Atlantic article on Google Mapping
Perhaps the most impressive thing is not that Google allocates human effort to the mapping project but that they combine very advanced algorithmics with that human effort and, perhaps, they know how to slide that along as technology advances.
I think a decent part of Microsoft's "Lost Decade" is that even though they have amazing talents at MSR, they were staffed for the 90s and the Web explosion and, with all that PC-focused staff, there were blind to or unable to shift towards the rapid emergence of a post-PC landscape.
If you look at things like handwriting and speech recognition, Microsoft had huge advantages in the early 00s (especially handwriting recognition: Dragon/Nuance has always seemed to lead in speech, but MS' handwriting recognition was (is?) miles ahead). Had MS invested in combining their technological lead with human-intensive fine-tuning in the same way that Google invests in map-making, Microsoft could be reaping the benefits today, instead of being roughly at parity.