Για να καταλάβετε πόσο επικίνδυνο είναι το Bitcoin Unlimited δείτε αυτό το σενάριο για το πως μπορεί το chain να γίνει συνεχόμενα hard fork και απλά το bitcoin να γίνει μπάχαλο.
https://www.reddit.com/r/btc/comments/5g1x44/bitcoin_unlimited_bu_median_value_of_miner_eb/το μόνο που έχει να κάνει ο επιτιθέμενος miner στο chain είναι να δημιουργήσει ένα block με μέγεθος την μέση τιμή έτσι όπως τα υπολογίζει το BU που δεν είναι και ότι ποιο απλό.
BU parameter data from last 2,000 blocks:
500 blocks - MG=2MB, EB=2MB, AD=4, Cumulative hashrate 25%
250 blocks - MG=2MB, EB=3MB, AD=6, Cumulative hashrate 37.5%
250 blocks - MG=2MB, EB=3MB, AD=25, Cumulative hashrate 50%
Possible malicious block size = 3.1MB
500 blocks - MG=2MB, EB=5MB, AD=3, Cumulative hashrate 75%
250 blocks - MG=2MB, EB=6MB, AD=16, Cumulative hashrate 87.5%
250 blocks - MG=2MB, EB=32MB, AD=2, Cumulative hashrate 100%
For any distribution of EB, there exists a median figure, which could split the hashrate.
για να καταλάβετε τι είναι το AD που είναι μέσα στους υπολογισμούς των τιμών γιατί και εγώ έσπαγα το μυαλό να δω τι στο διάλο είναι αυτό ο δημιουργός του Litecoin το εξηγεί καλύτερα
https://twitter.com/SatoshiLite/status/825063036767477760BU in NBA: Is 22ft 3pt distance too far for you? Shoot your 3pt shots at 21ft 4 times in a row, and refs with AD=4 will accept them as 3pts!
και εδώ εξηγείτε πολύ καλά
https://bitcoinmagazine.com/articles/closer-look-bitcoin-unlimiteds-configurable-block-size-proposal/Excessive Acceptance Depth, or “AD,” essentially overrules EB. More specifically, AD determines the number of added confirmations a block requires, before nodes and miners accept it regardless of that block’s size. The default is four.
So, let’s say a node sets EB to two megabytes, and AD to four added confirmations. If that node receives a three-megabyte block, it will initially ignore that block since it exceeds its two-megabyte EB. But if a majority of miners does not ignore that block, and mines four new blocks on top of it, the node’s two-megabyte EB is overruled by its four AD confirmations. The three-megabyte block is retroactively accepted as valid.
τώρα το σενάριο της επίθεσης ξεκινάει με την δημιουργία όπως είπαμε ενός block μέσης τιμής και συνεχίζεται με split συνεχόμενα του chain κατά αυτό τον τρόπο
New variant of the median EB attack
Step 1 - Launch the median EB attack. Successfully find just one block and cause a 50/50 hashrate split.
Step 2 - Hope the larger blocksize chain wins the battle, by triggering the AD values. Then 50% of "sticky gates" open, while 50% remain closed
Step 3 - The attacker now has 24 hours to make an even larger block, exploiting the fact that 50% of the sticky gates are open. If so, the network is split 50/50 again. (This is ironic since miners with a lower blocksize limit accept the larger block, and those with a higher limit reject it)
Step 4 - Hope the larger blocksize chain wins again. Now all sticky gates are open.
Step 5 - After 24 hours since the first set of sticky gates opened, they close. The hashrate can now be split 50/50 again. This process can continue
το άλλο που με προβλημάτισε είναι τι στο καλό είναι το sticky gates στο BU. Διάβασα αυτό εδώ
https://www.reddit.com/r/Bitcoin/comments/5h70s3/bitcoin_unlimited_bu_the_developers_have_realized/As part of the mechanism to accept larger blocks, Bitcoin Unlimited has a feature called a "sticky gate", when if a BU node sees a block greater than their local specified size "limit" (EB), and then this block receives AD confirmations, a “sticky gate” is opened for 24 hours (144 blocks). During this period the node apparently does not have any blocksize limit at all.
κατά την διαδικασία προσαρμογής του δικτύου στο νέο block size ανοίγει ένα sticky gate για 144 block όπου το block size απελευθερώνετε τελείως.
Κ.I.S.S (keep it simple stupid)
