In the Industry
For those who don't know, or don't have the position to know:
We use mainstream SSDs in our HP servers here. (well, half of them, we're transitioning slowly for obvious reasons). No SSD failures yet, and we've been running them for a year on our heavy-load production servers. The others are lighter-weight, so don't demand the I/O performance yet.
We've been "refreshing" our 3+yr desktops with little 50GB SSDs. Why not HDDs? Because we are replacing the disk drives currently in the machines as their performance tanks and internal error rates go up (which is usually a sign of immenent failure before SMART finds it). The machines run like night vs day, running the exact same image as the old machine had loaded. For the machines with <2GB of RAM, we drop in a bit of RAM and no need to refresh the hardware for another few years. Workers clamour for the old Core2duo (or a Pentium D in some [IT] humorous test-cases) rather than the newer quad-core 4GB+RAM machines. They just know "it's faster."
Failure rates among these desktop machines? None so far. Only been pushing them out for a year or so though (about when SSDs became viable price-wise for wide deployment). We now track new machines and have upgrade paths for them to get an SSD at their year-mark. Will we get failures? Sure. But we had terrible failure rates with the old spindle disks too (especially in our laptops).
So, you can argue about your MTBF assumptions, or reading a "4000 post forum!" which collects the disgruntled people who often repeat-post, and I'll continue deploying and being a satisfied SSD customer.