KenV, let's take a step back; remove the barbs from my previous post.
By focusing on just one item it is very difficult to avoid ending up blaming the pilot. The greater safety benefit is to explore why activities were not completed, or what factors might have contributed to an oversight, with the objective to aid crews alleviate the risks in these situations.
What has been published and discussed so far indicates that this was an 'unforeseen' accident; not an extreme unforeseeable event, but something which fell outside of the classic approach to safety. It appears to have emerged from a combination of factors involving many aspects of operation; these situations require a different approach to investigation and safety improvement.
For example, compare this incident with many successful GAs, where crew actions were correct, or if with deviation, were detected and corrected in time. Perhaps this accident involved an excessive variation in actions or corrective action took longer; the latter is more likely.
By considering the human as an asset in maintaining safety, as indicated by the successes of normal operation, then an output of accident investigation should identify aspects which increased the difficulty for the crew to act normally, and with this consider alleviation.
A major theme from the ASAGA report considered task saturation, workload, and time, against the backdrop of surprise.
Those aspect already discussed about systems integration, SOPs, and operational assumptions about new systems, should be reconsidered to reduce surprise, workload, etc, to effect change.
I refrain from choosing one item; emergent accidents require safety changes which involve many apparently insignificant areas of operations. Any one of which could prevented an accident, but without assurance for the next one - the next unforeseeable one; thus broad based changes are preferred.
We cannot continue to consider safety as good or bad, because modern complex operations require continuous adjustment to new (emergent) challenges e.g. the surprise of RAAS TOGA interaction. Safety involves continuous activity to manage the operational probabilities in risk, we attempt to keep the odds as low as possible.
"Can we prevent pilots from failing to add thrust during a go around? Absolutely yes."
I disagree; but we might reduce the probability, by reducing surprise, workload, systems complexity, and the proliferation of SOPs.
Help pilots detect and correct any oversight, to do what they already do well, but better.