Their drive buying habits have much in common with those found on this forum. They buy for TB/$, they will shuck external drives if needed, and from the look of their chassis design, they aren't afraid to run them hot. This is relevant to SOHO builds.
They reported a bathtub curve, which Google and CMU both rejected. I'd love to see their failure data separated by vintage, because they are showing much higher 0-2yr AFRs and much lower 2-4yr AFRs than previous studies. If the baseline has changed over the past decade, that's really good news for the short-warranty consumer drives, but with the reversal from previous studies and the disruption from the floods, I wouldn't jump to conclusions from the one graph they posted.
The long term annual failure rate numbers are the same as previous studies. By the fourth year, a drive's AFR is in the double digits and climbing fast. Drives just don't last.
They dinged Seagate on reliability in comparison to WD and Hitachi, but didn't go show the data. " Seagate – ST3000DM001‑9YN166 (slightly higher failure rate)". This correlates with what the warranty underwriters say, but seeing the numbers would be nice.
Despite their pre-deployment burn-in, a 5% AFR for individual drives means that 90% of their 4U pods see drive fail within the first year!. Two years ago, they had an employee spending one day each week replacing drives that have gone bad. Now that's either become a full-time job, or they switched to just leaving the dead drives in place like Google did.