The deal is, this is a flying computer, basically. Having built, supported and repaired computers for the last 20+ years, I have yet to see one that had 0 issues. Windows is updating constantly, trying to stay ahead of the problems. Why would this computer be any different? Firmware and Software updates are required to keep current with the system. With time, most all bugs will be found, if you have found one, it could be worth up to $30,000 to DJI, the reward for finding these bugs could be worth someone's while if they wished to pursue it.
But, talking about failures and such, we would need more information than a person's experience with purchases. I find it hard to believe that 7 out of 10 P4Ps have problems that require replacement. If that were the case, DJI would go bankrupt in a short time. Failure rates for products are 1-2% typical industry standard.
If these craft were that bad, they would not sell. Retailers would be overwhelmed with returns, DJI would find it impossible to keep up with the repairs.
Quoting a source "Gartner expects “personal” drone unit sales — mostly for photography, where DJI is strongest so far — to grow roughly 40 percent this year to 2.8 million, while it predicts “commercial” drone sales — with uses ranging from mapping to delivery — to grow about 60 percent to 174,000 in 2017."
Using those figures, that means they sold 1.68 million drones last year, 2.8 million this year. That is nearly 4.5 million drones it just two years. If your experience was typical, then of the 2.8 million sold this year, 1.96 million are flawed. With those kinds of returns, no company could stay in business. So, until DJI releases information on failure rates, which I doubt they ever will, I don't think there is as wide spread a problem as it may seem at times and we will never know the whole story. But, if there were that many failures, they would not be in business and having such success, don't you think?
I also wonder how DJI makes money. I've sold industrial computer boards to OEMs for over 20yrs too, and based on my experience, the reports I see from DJI craft, given the type of obvious flaws, this is totally unacceptable in the real world. If I was Frank Wang the QA manager would have been shot (I think you can do that in China

for such incompetency) a long time ago, allowing so many obvious defects to ship to customers. This is so easy to resolve, just implement outgoing inspection! That would certainly reduce many of the reported DOAs (in this case "Defective on Arrival). I have a hunch these
P4P's cost them about $200 to manufacture (craft+RC), that's how they make money. This company absolutely has a boatload of returns. If you remember in 2014 their repair dept took 8 to 12wks to get your craft fixed. They had to do something about that, so they implemented a "replacement" policy. I've noticed they rarely repair anything under warranty, I seem to always get a pristine new replacement craft, likely because they are overwhelmed with so many returns. This has happened 3 times to me in the last 18mos (2 X P4s and 1 Mavic) . That by itself implies the cost of the craft is less than the time they estimate it would take to diagnose and repair the problem that's been verified. Other pilots have reported getting refurbished units in exchange, because DJI USA rarely repairs anything. It also implies they don't have enough people and time to repair craft here in the US. Only once did they actually repair my craft, and that was in June when my
P4P had a bad gimbal out of the box (vibration that caused Jello). They actually replaced the camera/gimbal assembly, which takes very little time. In the last year I've had 4 warranty returns, and 3 out of 4 craft were replaced with new craft. How can they make money if I'm the average customer that's addicted to their technology?
The root problem is DJI doesn't have an outgoing test process for their craft before boxing them to ship. I'm not suggesting they fly each craft, but I'm suggesting they at least TURN IT ON and check for obvious things that are easy to spot, such as crooked cameras, or vibrating gimbal (I've have both of these happen to me), no gimbal obstruction messages, etc. If DJI would implement an outgoing test process, the number of flawed craft shipped would be reduced dramatically. Just boot up with each battery to insure it can be charged would be helpful. There are tons of reports of DOA P4 batteries. That's absolutely disgraceful that so many batteries fail to take a charge. This rarely occurred with P3 batteries.
And yes, today it's likely that 95% of all
P4P's cannot record video at 4096X2160X60FPS as advertised, without glitches. 3860X2160X60FPS works fine. Same with gimbal flips, I estimate over 75% of all P4Ps can reproduce a gimbal flip simply by flying full throttle forward (40+mph) in S mode and turn left. Mine does it, and I think everyone I know with
P4P can make theirs do it. Some do this easier than others, but most all craft exhibit this anomaly. If you use P mode with OA enabled to restrict speeds to about 30mph you won't see this, unless you're in 15+mph winds, which has proven to increase occurrences.