You seem to assume here again, at least during these few paragraphs, that the attackers try to steal a larger share of the newly minted coins (as miners). This is not their objective. Their objective is to steal a lot more than that via trades that they then rewrite afterwards, keeping only their ingoing transactions. (And their ultimate objective is actually to cause a crash of the cryptocurrency, assuming that this is what will happen.)
if they keep only their own tx. they cant do double spends..
remember they can only double spend their own value they control so if they are not reversing their own transactions then they cant double spend
also by reversing other peoples transactions they cant then take control of other peoples. because the malicious side does not have the key to sign the fund of other people. so in no way can a malicious side steal funds by reversing other peoples transactions.
emphasis a double spend is only able to happen by the malicious side reversing its own transaction and then re-spending their own value
to which i explained to achieve that the people in the attack need to convert their btc to goods/services/other currency. receive those goods/services/other currency in a settled final manner.. and THEN they can undo the blocks containing the transactions they want to reverse knowing they already have the value settled in another form(goods/fiat/altcoin). to then know when the btc transaction is reversed they can then respend the btc again to a different recipient.. to double spend that same btc
[...]
as for you saying about re-orging blocks to then mess with the market to crash it.
to be successful with that when the malicious pools deposits coin into an exchange. it has to wait for the exchange to deem the funds are settled with them (6 confirms for any significant amount deposited) the malicious users would then have to waste their deposit balance on orders to force a crash. and then remove the other currency to then re-org the blockchain to undo the deposit. so they can then do it again
however exchanges will notice these tactics of re-using a utxo thats was previously spent, and just block/ban users thus avoiding users abusing their balance database and market orders
also as said to do this, they cant just respend the same utxo every block by doing re-orgs every block. as i explained they would need to go through a process of delaying a 51% blockchain attack by ~50-4000 blocks to play each round out to then re-do it again.. by which time even the invited people to the malicious pool whom bought into bitcoin hardware will see the negative affects actually hurt their investment and they can within seconds jump to honest pools. where by it takes a malicious pool multiple hours/days per attack round
however in a ethereum attack the custodian of stake can manipulate blocks whereby the stakers wont counter it, because the stakers funds are at risk(the penalty) and the stakers cant simply jump to a honest custodian in seconds because it takes a day to de-stake. so the risk of a dishonest custodian on ethereum is far more harmful to ethereum users than a bitcoin attack is to bitcoin
if you want to delve dep into a blockchain attack to effect a CEX market you really need to learn the difference between the blockchain transactions which are not the market price orders vs the CEX balance and market order databases which are not the blockchain
then run scenarios on whats actually involved in doing an attack and how things operate
dont just side step things simply because it doesnt appease you hopes that people simply dont say ethereum is king
instead learn the mitigating factors of reality and realise bitcoin has alot more strengths than ethereum does
like i said if you are attempting to re-org the blockchain in a 51% attack to double spend funds to continually crash the market. you need to learn how the delay of the confirms. the length of "catch up" time and also the mitigating factors a CEX can put inplace in regards to its balance and market order database and services decisions all are factors
its very easy for a CEX to keep a log of the UTXO's being spend as deposits. and then ban users that try re-using the same UTXO even in a block -reorg situation
You seem to assume here that the 51%-attackers needs to make only one replay per reorg. But in fact they can make several replays per reorg:
Suppose Alice trades 1 bitcoin with Bob for some tokens or some USD, then trades that for "another" bitcoin from Claire (meaning that Claire's ownership of the coin isn't dependent on the first transaction with Bob), then trades that bitcoin away again to Doris, then buys "another" bitcoin from Eric. And suppose that Alice is then able to rewrite this recent part of the ledger afterwards. Then Alice can keep the transactions with Claire and Eric, i.e. where a bitcoin is transferred to a wallet of Alice's, but replace the transactions with Bob and Doris with two other transactions where the bitcoins are instead transferred to two other wallets of Alice's. At the end of this, she will have 3 bitcoin in 3 separate wallets: the one she started with and the ones from Claire and Eric.
And she could in principle have kept repeating this process (before rewriting the ledger) as many times as she can find traders whose ownership over the traded bitcoin isn't dependent on earlier trades with herself (i.e. she can only replay each single bitcoin once).
Now turn this example into Alice instead being a great number of people, who are backed by billions of dollars in total to do this attack.
And furthermore consider the fact that it is typical to see around $15B being traded each
day. (And again, you agreed that Ethereum investors could in theory afford an attack lasting for several
months, once they've paid the CapEx.)
And like I've said: the confirmation period unfortunately cannot be changed retrospectively, at least not with pure PoW.