builder0x69.eth

@ballsyalchemist I don't understand why it's surprise for anyone. After low-card-crusader all relays already got stricter on timing and reject late block payload requests. Forked blocks still happen from time to time even if less frequently for mev-boost than classic: etherscan.io/blocks_forked

Slow relay leads to private txs being DOXXed I think this could open up a can of worms for all sorts of "interesting" ways to exploit private txs. Relay gets bribed to delay certain blocks, two forks get into contention...

@ballsyalchemist @0x94305 Correct In most cases participation is 99% or higher so majority of nodes agree. But in cases where block was released very close to cut-off(3 seconds into next slot) you can have split between nodes so some will reject block received late and consider all transactions as public

@builder0x69 @0x94305 Just to clarify, you are saying that all txs become publicly available as in the block becomes public as per usual regardless of the final voting weight. And then the proposer Boost can kick in finalize that block, resolving the contention between the forks. Is that right?

Turns out this is a massively underrated tweet. No, I will not elaborate.

@0x94305 Re-phrasing my answer to be more polite: 1) Mempool guru data is wrong and all 8 TXs were private in this block. 2) Likely their node received 17670167 late and didn't consider it canonical. 3) Next slot had only 47% sync participation but won anyway with proposer boost.

@0xdapper_ @hrkrshnn We do support gas-less bundles so EOA can have 0 balance initially, but constructing a bundle is a rather hard task for normal users. Also most popular wallets won't allow you to broadcast transaction from empty wallet even if using custom RPC provider which accepts them.

@hrkrshnn Can kinda still get 'clean ETH' if you use @builder0x69 or similar services that'll front your bundle gas fees as long as you return it within the bundle

An unfortunate downside of 1559 and the merge is that it killed the idea of 'clean ETH'. Pre-1559, one can get 'clean ETH' by 1. finding an arbitrage in the network, 2. write a contract that executes it in the constructor, 3. forward a portion of the arbitrage to the miner, 4. forward the remaining to a fresh address of your choice, 5. sign and send this tx to a flashbots relay with `gasprice = 0`. Now you have an address with ETH that's completely unlinked. The ETH comes from proof of arbitrage, i.e., proof of work! Before the merge, you could perform another kind of work, with GPUs and joining a mining pool. The ETH that you got from it was even cleaner! I'd love to see EIP-1559 loosened up so that the arbitrage trick can still work.

@MevRefund @sensei_dot_win Per documentation it should be 40% user, 15% bloxroute, here it looks like 55% bloxroute in case of etherscan.io/tx/0x023ad8874… Though may be indeed it's not backrunme endpoint but something else.

@builder0x69 @sensei_dot_win But where's the user refund? I only see bloxroute and bot payments.

Speaking of better, mind explaining how bloxroute is monetizing their private flow? Just discovered that this address etherscan.io/address/0x965d… belongs to bloxroute. Appears to be profiting from private order flow without offering any user refunds ...

@MevRefund @sensei_dot_win At the same time I think users do get refunds, just access for searchers is permission-ed(unlike other OFAs) so searchers make the payment. Per distribution in: docs.bloxroute.com/introduction/b… Still sharing transactions with permissioned group of paying searchers is not same as private.

@MevRefund I can tell you more, recently @sensei_dot_win reached out about weird behaviour for bundles they submit. It looks like bloxroute automatically add refundPercent/refundRecipient to bundles they forward to other builders like us and that's why you see so many refunds from us.

@MevRefund Also want to clarify that automatic gas-less bundles were advertised way back in November: @builder0x69/50-000-blocks-and-small-new-feature-31b1d67f6fb9" target="_blank" rel="nofollow noopener">medium.com/@builder0x69/5… It means that builder will automatically fund any EOA inside the bundle if it has insufficient funds.

The searcher had an opp tx and a bribe tx, and was relying on the bundle only landing if the bribe EOA won enough money from the opp to bribe correctly. However, builder69 is happy to chip in if they end up making more money. Searcher ends up bribing 1 Eth for nothing!

Found a fun bundle from a few months ago which can be summarized as: builder69: I see you missed out on that 16 Eth opp. Better luck next time! Anyways, here's some money so you can still bribe me! searcher: 😢

@MevRefund @0x94305 That was advertised way back in November: @builder0x69/50-000-blocks-and-small-new-feature-31b1d67f6fb9" target="_blank" rel="nofollow noopener">medium.com/@builder0x69/5… And gas-less bundles are beneficial for 99.9% of bundles, that's indeed a weird edge case since searcher decided to ignore main principle of bundles that transaction should revert in case of failed opportunity.

@0x94305 @builder0x69 It's not a coincidence. builder69 has advertised that they will automatically fund EOAs (i.e. gasless bundles) if it's profitable to do so. Poor searcher just found out the hard way.

@0x94305 @0xMikolaj @should_b_workin @NatashaVasan But this discussion already turned out too long for twitter shit-posting so I'll stop.

@0x94305 @0xMikolaj @should_b_workin @NatashaVasan Well then just create bait TX moving price a lot on some ill-liquid yet popular ERC20. For your (1) ETH-USDC example they don't need any private information whatsoever since they can just always bid 100% in their own blocks.

1/ It's very simple. Integrated searcher-builders who get private order flow are risking ending up (unknowingly) insider trading. Tools that promise front-running protection or privacy shouldn't send transactions to such builders. Other builders should publicly commit not to..

@TheDEFIac @0x94305 @0xMikolaj @should_b_workin @NatashaVasan You mean in both arms of A/B test you didn't get any fills? Well if you take a week and split it into even/odd blocks, half of good opportunities will be in even blocks, half in odd blocks. But if you don't get any fills in both arms of test you're probably just a bad searcher?

@builder0x69 @0x94305 @0xMikolaj @should_b_workin @NatashaVasan Yes, but then what are you trying to test with such A/B test if you won't get any good fills in A nor B? You want to know whether you get filled on the good opportunities and for that block number is just not enough.

@TheDEFIac @0x94305 @0xMikolaj @should_b_workin @NatashaVasan Which is true in both cases of 50%/50% split, no? Or just measure your adverse selection then during the A/B test

@builder0x69 @0x94305 @0xMikolaj @should_b_workin @NatashaVasan Bc you might get filled only in situations when price moves against you between time at sent and time at block created.

@TheDEFIac @0x94305 @0xMikolaj @should_b_workin @NatashaVasan My example was mainly for (2) copy atomic arbs where I think no-ones cares about nature of ERC20 being traded. For non-atomic trading probably it would indeed require some capital to setup the bait/test.

@builder0x69 @0x94305 @0xMikolaj @should_b_workin @NatashaVasan I think the point is whether this happens on the pools the builder integrated searcher is already compeeting on.

@TheDEFIac @0x94305 @0xMikolaj @should_b_workin @NatashaVasan What's the problem with targeting just a block number? Over longer time period with 50% split it, price spikes should even each other out.

@builder0x69 @0x94305 @0xMikolaj @should_b_workin @NatashaVasan How would the A/B test look like? You cannot just target a block number, bc you will get filled when price moves against you and won't when it doesn't. Also fairly confident that there is less than 10 searchers atm able to do this + it will take a ton of time.

@0x94305 @0xMikolaj @should_b_workin @NatashaVasan > How do you design honeypot txs? Just create a new honeypot contract/token no-one knows about it so probability of trading with it / using it is 0 unless your transaction was copied. Many such examples to bait in the mem-pool.

None of this proves anything useful, unfortunately. Say I’m a builder who decides to (1) outbid everyone else on USDC-ETH and similar arbs (as long as they are profitable), and (2) copy atomic arbs. How do you design honeypot txs? On the second one, ok, you learned that your inclusion rate with me is lower than with Other Builder. I’ll say, sure, we are a smaller builder specializing in stat arb and simple atomic. We can only win when tx of types (1) and (2) above are competitive. Your txs are not. Other Builder can still win with your lousy txs because they are big and handsome, but we can’t. And by the way, I have heard from searchers that they exclude builders based exactly on such suspicions but nobody can ever prove anything. So I think the only way to proceed is to make sure that when “private transactions” are advertised to the general public, they are not sent to searcher-builders.

@TheDEFIac @0x94305 @0xMikolaj @should_b_workin @NatashaVasan Any large searcher or large OFA should be able to get statistically significant results rather quickly. Tech for 50% A/B test based on block number doesn't sound like rocket science. In terms of capital as a searcher/OFA you're already sending these bundles anyway?

@builder0x69 @0x94305 @0xMikolaj @should_b_workin @NatashaVasan That would take quite some time to be able to collect any meaningful amount of data. Also has quite high requirements for your capital and tech to do this.

@0x94305 @0xMikolaj @should_b_workin @NatashaVasan If you're a searcher you can also run basic A/B test: 1) 50% of blocks you send only to builders you highlighted above 2) 50% of blocks you send to all large builders(even searcher-builders) Would love to see results in terms of captured opportunities / profitability in general.

@0x94305 @0xMikolaj @should_b_workin @NatashaVasan Just send bait/honeypot transactions as an outsider? As searcher you can also easily start sending your bundles only to subset of builders while measuring percent of blocks where someone paid more for the same opportunity as you.

@0x94305 @0xMikolaj @should_b_workin @NatashaVasan There're lots of stats you can compute, information whether TX was public or not is also available for everyone. As searcher you can also exclude builders 1-by-1 when sending private txs and compare results. But like any proper analysis it takes time so way easier to speculate.

As I pointed out right in the first tweet, the issue is that insider trading might very well be *unintentional*. TradFi broker-dealers invest lots of money in controls and then still end up paying fines for insider trading, front running and trading ahead. May be mistaken but I doubt that integrated searcher-builders have compliant controls. Lastly, from the DeFi perspective, the issue with some of the situations described above and elsewhere is that they are inherently unobservable to an outsider. You submitted a private tx with a juicy arb but it didn’t make it. You see a competing one in the winning block. Can you tell if it was designed based on your tx?