CMV: The main Israeli criticism of the NAZA documentary is hollow distraction
CMV: The main Israeli criticism of NAZA seems pretty hollow
For anyone who hasn't seen it: NAZA is an Israeli documentary based on testimony about IDF targeting in Gaza, including AI generated targets and how many civilian deaths could be accepted for individual targets.
The movie has 24 anonymous IDF/intelligence sources on screen, but the investigation apparently involved more than 100 sources, including senior officers and people involved at different stages of targeting/approval.
The Guardian reports that claims were cross-checked against other sources, specific strikes and classified material.
And the overall picture isn't exactly coming out of nowhere. UN investigations, Amnesty and other organisations have separately documented huge civilian harm, destruction of civilian infrastructure and extremely permissive Israeli conduct in Gaza.
Yet I keep seeing basically the same arguments against the movie:
"The witnesses are anonymous."
Anonymous to us.
The journalists know who they are and checked their stories against other sources/evidence.
Do people seriously expect serving IDF personnel accusing their own military of war crimes to appear with name, rank and address?
"They are just junior soldiers who don't understand the targeting process."
Except the investigation included more than the 24 people shown and reportedly included senior officers and people directly involved in targeting and authorization.
So this seems to be arguing against something the movie isn't actually claiming.
"There is no evidence. Just testimony."
Again, no.
The investigation reportedly used 100+ sources, classified documents and checked specific attacks.
You can argue that they should release more of the underlying evidence. That is reasonable.
But that isn't the same as saying it doesn't exist.
"No Israeli strike killed 500 civilians, therefore the claim that 500 civilians could be accepted is false."
This one makes no sense to me.
The allegation is about the estimated civilian deaths BEFORE the strike.
If you authorize a strike expected to potentially kill 500 civilians and it ends up killing 80, that doesn't magically mean the original authorization didn't happen.
"Those numbers weren't allowed civilian deaths. They were authorization thresholds."
Ok... but what happens when it reaches the person with authority to approve it? That is the relevant question.
If 20 deaths needs approval from person A and 100 from person B, saying "100 wasn't automatically allowed" doesn't answer whether person B then approved it. Nor does it provide statistics on how many times humans approved or disapproved the targets or what the process was.
It does not help when sources claim:
>B. said that the reason for this automation was a constant push to generate more targets for assassination. “In a day without targets [whose feature rating was sufficient to authorize a strike], we attacked at a lower threshold. We were constantly being pressured: ‘Bring us more targets.’ They really shouted at us. We finished [killing] our targets very quickly.”
"Lavender isn't an AI kill list because a human approves the target."
Again this seems mostly semantic. The allegation is that algorithms generated enormous numbers of targets and human review became extremely limited.
If a person looks at an AI generated target for 20 seconds and clicks approve, technically a human made the decision.
That doesn't answer the criticism. Especially not when it is a heuristics where the score needed to be deemed "likely hamas operative" can be changed if they do not get enough targets.
"Israel calculated civilian casualties, therefore they were trying to protect civilians."
Calculating civilian casualties obviously can be part of protecting civilians.
But surely the interesting part is what number you then ACCEPT?
A system saying "this strike will probably kill 80 civilians" doesn't become humanitarian because someone calculated the 80 first.
"Hamas operates among civilians."
Yes. That makes targeting harder.
How does it answer whether Israel massively loosened its acceptable civilian casualty limits or whether targets received meaningful review?
Additionally the claim is that the identification of "hamas"-people is 85% precise, though that number is a claim, not a fact. But 15 innocents + families out of a 100 targets is very "non-precise targeting" for precise strikes.
The logic seems to be that all people targeted are Hamas despite of this.
"They ignore October 7."
October 7 explains why Israel went to war. It doesn't give Israel unlimited freedom in how it conducts that war. (and time did not start on October 7)
"The filmmakers are anti-Israel."
Maybe. Who cares? Are they Anti-"the right wing government" (= 75% of elected)
Show that their evidence is wrong.
The thing I find interesting is that most of these arguments don't really address the central claims.
The relevant questions are:
- Did Israel substantially increase how many civilian deaths it would accept for targets?
- Were huge numbers of targets generated algorithmically?
- Did those targets get meaningful human review?
- Were strikes approved while knowing large numbers of civilians would probably die?
If NAZA is wrong, those are the claims I would expect the rebuttal to disprove.
Instead most of what I see is "anonymous sources", arguments about terminology, attacks on the filmmakers and distinctions that don't actually contradict what is being alleged.
CMV: What is the strongest actual evidence that the central claims in NAZA are false?