After reading thousands of posts about this, I guess my conclusions are:
1. Most (bit not all) people are reasonably confident in the independence and thoroughnes of the investigative process, and the conclusions drawn. Many were surprised they managed to recover the recorders, and then the data from the recorders. Perhaps as a result of previous criticism, BEA have made greater efforts to be transparent this time around, and recorded crucial steps (such as the sealing of the recorders). They recovered the two bodies to establish with 100% confidence who was seated at each position.
2. Apart from the pitot icing (a known problem), the aircraft most likely operated as designed. The UAS by itself should not have been sufficient to lead to this tragedy, and numerous previous incidents ended uneventfully.
3. There may be some aircraft design features which need to be reviewed. The stall warning cut out at low speed does seem unnecessarily confusing and may have been unfortunate in this case. There may well be others, such as the sidesticks being out of view of the other pilot.
4. I have read a lot of discussion about various control laws, and am frankly surprised at how much disagreement there is. Have things become too complex? Can pilots be expected to work through this in real time if these issues can be debated for many months here?
5. There should also be some thought given to the human - machine interface when things go wrong. For many reasons, the trend towards increased automation will continue, and pilots of the future will have less real "flying" experience. It is inescapable that these developments create a challenge when the automation fails. This challenge is predicable, and must be addressed. In this case, the pilots were overwhelmed by warnings and messages and conflicting data, and this contributed to the poor situational awareness and loss of control. This is also predictable. In an emergency, information overload will quickly lead to a feeling of helplessness and panic, and also confirmation bias as you search for something (anything) that fits with your (potentially erroneous) understanding of the situation.
6. There are obviously CRM considerations. After the upset obviously. But also prior to the upset. Was a 40 second handover sufficient before the Captain left the deck? At night, crossing ITCZ? Is it reasonable that the Captain suggested that a pilot had a sleep after he (Captain) went for his rest?
7. Video recording would greatly add to the CVR in the event of an incident. I don't believe this should be released to the public - some things do belong behind closed doors. But a visual record would surely strengthen the investigators understanding of exactly what happened, would lead to firmer conclusions, and would enhance lessons learned. In addition, the knowledge that there was a visual record available would help improve confidence in the investigation - in this case, we would know that the displays were functioning. Solid state data storage is cheap, compact and secure. Given that I am recorded when I scan goods through the supermarket checkout or pull out $50 from at ATM or fill my car at the pump, then maybe this is a development whose time has come.