EPTThere are some pretty obvious practical and epistemic problems.
Take your example of a unit saying, "We came under fire," or "We couldn't identify the vehicle as an ambulance." Why should an investigation automatically assume that this account requires an extensive independent reconstruction?
You could send investigators to the site, interview witnesses, examine footage, communications, weapons, damage patterns, etc. But if the relevant evidence is gone, you're eventually left with competing accounts. The absence of evidence that the unit was fired upon doesn't establish that they weren't fired upon, just as a civilian witness saying they weren't under fire doesn't establish that there wasn't a shooter somewhere else.
And that's where the resource issue becomes relevant. A genuinely thorough investigation can consume a considerable amount of personnel, intelligence resources, forensic work, interviews, reconstruction, etc. If you have thousands of incidents, you have to decide which ones justify that level of investigation. You can't realistically reconstruct every disputed engagement to the same standard.
There's also an important distinction between "the IDF has an incentive to investigate" and "the IDF has the same incentive to investigate as the people affected by the incident" or "Hamas has an incentive to investigate to damage the IDF's reputation".
The IDF obviously has reasons to investigate its own mistakes: operational lessons, discipline, legal exposure, protection of its soldiers, and so on. But it also has an institutional interest in defending its personnel and operations. Gaza's civilian population, victims' families, journalists, aid organisations, etc. have a very different set of incentives. Hamas as well.
That doesn't mean the IDF's account should automatically be trusted / distrusted. It means that there is a legitimate reason for some cases to be independently scrutinized, particularly where there is substantial evidence suggesting that something went wrong.
I'd therefore distinguish between:
1. routine fact-finding where the available evidence is sufficient to establish what happened
2. more extensive military investigation where there are credible indications of misconduct or a serious procedural failure
3. genuinely independent or criminal investigations where there are substantial indications of serious wrongdoing
I don't think every disputed incident necessarily deserves a full independent reconstruction simply because somebody contests the military's account. Sometimes the evidence will simply be insufficient to establish what happened, and no amount of post-war investigation can change that.
So I don't think the disagreement is really "investigate vs. don't investigate." It's more about thresholds, resources, independence, and what level of evidence can reasonably be expected to establish the truth in a chaotic war zone.
That was exactly my point about data: If we don't know the denominator, don't know how many disputed incidents have legitimate explanations, and can't reliably reconstruct many of them, then we should be very cautious about drawing broad conclusions from the subset of cases that become publicly controversial.
We are now looking at one of the accounts where the IDF decision was wrong. But how many are not? How does that ratio fare against other war zones that weren't nearly as heavily documented? Where the enemy didn't use such extreme underhanded tactics or didn't distinguish between civilians and fighters? It is an impossible task to solve for us but you are right, that the IDF should - to the best of its abilities and resoures - bring to surface what really happened.
