Gambit Labs

@0xe8_team @MevRefund DM sent on Discord. Truthfully missed your first reply in the thread. Saw the this one and then your message on Discord. Let’s continue the conversation.

@gmbitco @MevRefund update: we tried to contact Gambit Labs in different ways. but they never responded.

Yikes, Mev Bot 0xe8 just got unbundled, losing 57 Eth.

@MevRefund (Added additional checks/guards and monitoring to handle this case; assuming it is what in fact happened)

@MevRefund After building that block, payout was not competitive enough so we switch to giving all proceeds to the proposer by setting the coinbase to the proposer address. We do it in a fast rebuild, without additional revert check. I suspect that’s what happened here. Still investigating.

@MevRefund See next comment :) tweet got too long.

@gmbitco 1) What So it only reverts on Etherscan?

@MevRefund Bit more info, the revert was due to out of gas trying to pay the coinbase (which was set to the proposer). Here’s the trace etherscan.io/vmtrace?txhash… (scroll to the end). The proposer is a contract (does fee distribution etherscan.io/address/0x2196…)

@MevRefund We don’t unbundle any requests sent to us, and we drop reverting bundles while respecting the reverting tx hashes field. Also, the coinbase was set to the proposer, we didn’t take any profits in this block. Will investigate further why it reverted.

@MevRefund Agree, wouldn’t make sense for them to do that.

@titanbuilderxyz ^ in case someone running a relay is interested in further exploring from that end (@ultrasoundmoney, @blocknative, @GnosisDAO)

@titanbuilderxyz These are their pubkeys (as seen from last 50 blocks): "0x82790923494535e417675b13e580be40d15bd3a8ddcab925e6c19e919c55b6e7bed5037925ba67caeb715dcc74d720da" "0xa4024c0050986bceee0b58e24455c7bb620a6ca96725026f315a6e3b4114fdc54092c4c87f6de72a39413371a27e5d83"

📢 PSA: We have identified some unusual activity regarding a builder building blocks using our extra data "Titan (titanbuilder.xyz)". This is what we know so far 👇

@SheaKetsdever To be clear, * Data is available with a delay of N blocks * Shared is: bundle hash, bid amount, and if it reverted * Data is only queryable via a target tx hash What issues does FB see with this level of visibility that goes against FMP? Is there specific data that's of concern?

@SheaKetsdever Would appreciate more detail/explanation for the specifics we're proposing here :) It's important to make the distinction of what's being revealed. We're thinking of this in terms of there being a minimal amount of bid info that benefits searchers/users without causing any harm.

We’re upgrading Flashbots Protect to make private transactions faster⚡️ You can now connect to more builders through the Protect RPC docs.flashbots.net/flashbots-prot…

@SheaKetsdever We are experimenting with this view into the auction helping searchers understand where their bids sit against competition or to debug their systems (gmbit.co/auction). Bundle content is not revealed. Is this problematic for FB concerning the principles outlined above?

@SheaKetsdever How does FB think about privacy concerning past blocks and anonymization? And in particular revealing data about bids vs bundles themselves (like the relay data APIs).

We also built a light visualization on-top of this new API. Here's an example gmbit.co/auction?txHash…

We intend for these APIs to aid searchers while maintaining privacy through anonymized fields, delayed results, and aggregation. You can read more at gmbit.co/auction and start putting it to use today with our Block Builder. Bundles you send you can then query and compare.

Don't search blind. With gmbit.co/auction Searchers now get visibility into the bundle auction. We're proposing APIs served by Block Builders to help Searchers understand how intense competition is around a target tx, check overbidding, and if the builder is rational.