Flashbots(blog.bitmex.com)
blog.bitmex.com
Flashbots
https://blog.bitmex.com/flashbots/
66 comments
Semi-aside: the term “front running” is pretty overloaded and more than a little charged at this point. It gets applied to a whole spectrum of activities from the socially positive (racing to the inside with a post-only limit order to minimize the still positive price improvement) to the murky and probably kind of shady (MEV ecosystem) to outright fraud (an agency execution firm literally loading up before executing a client’s block, intentionally mispricing an IPO).
I would submit that it’s probably best just avoided by anyone who’s objective is clear communication. It’s still great for ruckus-making, but that’s probably more of a bug than a feature on HN.
FWIW I’m also very skeptical of modern MEV-world. I just think “front running” is a bad term for anything these days.
I would submit that it’s probably best just avoided by anyone who’s objective is clear communication. It’s still great for ruckus-making, but that’s probably more of a bug than a feature on HN.
FWIW I’m also very skeptical of modern MEV-world. I just think “front running” is a bad term for anything these days.
It just means detecting a transaction before it happens and executing it earlier for monetary benefit.
In a PoW cryptocurrency blockchain context, maybe? Though I think the much more precise term MEV captures that nicely.
The etymology of the term "front-running" goes at least as far back as the name "Reuters" and the use of pigeons to be the microwave towers of the day: https://www.reuters.com/article/rpb-historyofspeed/the-long-..., and probably a great deal further. I imagine the actual origin of the term is lost in deep antiquity due to the fact that someone with tradable information in hand would profit more by "running" than by walking.
I don't personally deal for profit in mempool analysis, but I'm the "low-latency guy" to enough people that do that I've had to learn the mechanics of a few of the big chains, which is frankly time that I'll never get back, and while the Nash equilibria are far less stable/obvious in this world than in a good old fashioned price-time precedence order-book, it looks pretty fucking shady to me.
The etymology of the term "front-running" goes at least as far back as the name "Reuters" and the use of pigeons to be the microwave towers of the day: https://www.reuters.com/article/rpb-historyofspeed/the-long-..., and probably a great deal further. I imagine the actual origin of the term is lost in deep antiquity due to the fact that someone with tradable information in hand would profit more by "running" than by walking.
I don't personally deal for profit in mempool analysis, but I'm the "low-latency guy" to enough people that do that I've had to learn the mechanics of a few of the big chains, which is frankly time that I'll never get back, and while the Nash equilibria are far less stable/obvious in this world than in a good old fashioned price-time precedence order-book, it looks pretty fucking shady to me.
It is used as a trading term in hedge funds. It is shady and heavily regulated by SEC unless you are crazy enough to trade crypto or fx where SEC can't regulate anything.
Came here to write the same. I'm real happy when someone runs ahead of my limit orders and provides me liquidity/execution. If I wanted to pay less, I would have set the limit lower.
You can set a slippage tolerance when you make an order on a decentralized exchange. It's basically the same as a limit order priced to market, where you effectively set the "most you're willing to buy for" / "least you're willing to sell for" and occasionally get better rates.
With a DEX your transaction might just not execute at all some of the time of creating the transaction, so you might burn a transaction, and there's no orderbook for the order to sit in til later.
With a DEX your transaction might just not execute at all some of the time of creating the transaction, so you might burn a transaction, and there's no orderbook for the order to sit in til later.
What you call socially positive others may call execution cost. Because you might improve on a pension fund limit order. So it's best to not get into this territory.
A non-aggressing order (I'll use post-only limit as an example, but as you probably know there are a zillion kinds of PoL and IOC and flags and stuff that meet this criteria) strictly makes the book deeper, or narrower, or both.
A pension fund (or rather the agency execution-type firm that they'd be crazy not to employ for block trades) wants deep books and narrow spreads: if they were in the business of collecting maker rebates then they'd be market makers, which would make them terrible pension fund managers.
If their agency execution firm is able to shave a bit of slippage/friction off by leaving some passive orders in the book I suppose they might have some luck with that, but there's no way that punishing retail traders with wider spreads on the off chance that an agency execution firm has to have a socially-approved client is a net win under any definition of socially positive. Execution cost for non-advanced actors is wide spreads, shallow insides, and high fees: strict market makers improve all of that.
People who make arguments like this one are usually day traders with no business using limit orders. I'm not saying that's what you are, but that's usually where you hear it: certainly to the extent that this argument has an intellectual pedigree other than that awful Lewis book, that's it.
A pension fund (or rather the agency execution-type firm that they'd be crazy not to employ for block trades) wants deep books and narrow spreads: if they were in the business of collecting maker rebates then they'd be market makers, which would make them terrible pension fund managers.
If their agency execution firm is able to shave a bit of slippage/friction off by leaving some passive orders in the book I suppose they might have some luck with that, but there's no way that punishing retail traders with wider spreads on the off chance that an agency execution firm has to have a socially-approved client is a net win under any definition of socially positive. Execution cost for non-advanced actors is wide spreads, shallow insides, and high fees: strict market makers improve all of that.
People who make arguments like this one are usually day traders with no business using limit orders. I'm not saying that's what you are, but that's usually where you hear it: certainly to the extent that this argument has an intellectual pedigree other than that awful Lewis book, that's it.
Another semi-aside: The term "censor transactions" needs to be abandoned, because it accepts the whole "money is speech" nonsense at face value.
I'm willing to begrudgingly accept "money spent to effect speech is speech" but no further. The vast majority of crypto transactions do not fall within that narrow bucket, and using the word "censor" to refer to reversing them with a 51% attack is propaganda.
I'm willing to begrudgingly accept "money spent to effect speech is speech" but no further. The vast majority of crypto transactions do not fall within that narrow bucket, and using the word "censor" to refer to reversing them with a 51% attack is propaganda.
My understanding is that when people say Money is speech what they really mean is that it is a human right.
When an authority freezes your bank accounts (like Trudeau/Freeland did to the truckers) they infringe on your other rights (right to property, right to shelter, right to assembly , etc) demonstrating how much of a slave people are under these regimes.
When a payment processor cuts you off without recourse, they also infringe on your rights.
Money that most people know today is a terrible way to store value as it is also a tool for the state to control them.
Ethereum is now another government tool since a great percentage of transactions are validated within the US jurisdiction and are thus subject to the same draconian laws as federal bank issued money.
When an authority freezes your bank accounts (like Trudeau/Freeland did to the truckers) they infringe on your other rights (right to property, right to shelter, right to assembly , etc) demonstrating how much of a slave people are under these regimes.
When a payment processor cuts you off without recourse, they also infringe on your rights.
Money that most people know today is a terrible way to store value as it is also a tool for the state to control them.
Ethereum is now another government tool since a great percentage of transactions are validated within the US jurisdiction and are thus subject to the same draconian laws as federal bank issued money.
You're arguing against basic human rights.
While technically correct, this post is such a gaslight exercise. It boils down to: Frontrunning is profitable and bad for the system. So let's discuss our totally trusted system optimising the frontrunning. It's good for everyone, promise. Just don't ask about the impact on the transaction throughput and fees that comes from those transactions existing in the first place. And don't google rent seeking. Remember optimising a bad thing is good, because it's a local minimum, so let's make sure to restore it in eth2!
> In Ethereum, as of May 2022, it is believed that over 90% of miners are connected to the Flashbots server. One can determine this by observing that over 90% of blocks contain Flashbot transaction bundles.
> The Flashbots system is integrated into the current Proof of Work (PoW) mining system on Ethereum. When the Merge happens and PoW mining is switched off, Flashbots adoption will reset to zero again.
This article is kind of dated at this point.
> The Flashbots system is integrated into the current Proof of Work (PoW) mining system on Ethereum. When the Merge happens and PoW mining is switched off, Flashbots adoption will reset to zero again.
This article is kind of dated at this point.
> We go on to look at ETH 2.0 and explain why Flashbots or a similar system needs to be constructed again from scratch.
I read it as what was and now what is needed
I read it as what was and now what is needed
Mev-boost, a piece of software that consensus clients delegate their block building to, is currently being used on mainnet. The difference is that there can be multiple block builder _relays_, such as Flashbots, bloXroute, Manifold Finance, and Blocknative.
These relays effectively accept bundles from MEV searchers, which are then sent to the validator running mev-boost as they propose their block.
https://boost.flashbots.net/
These relays effectively accept bundles from MEV searchers, which are then sent to the validator running mev-boost as they propose their block.
https://boost.flashbots.net/
The title might be misleading to those not well versed with how block production works.
> [flashbots] is not, strictly speaking, a necessary component of the block production process. The centralised Flashbots relay entity does not directly own any hashpower. Nor can Flashbots censor non-MEV related transactions. Miners can still include transactions that did not go through the Flashbots relay.
This is also likely to change as MEV Boost relay code was recently open sourced to accelerate and promote more relay diversity.
https://www.theblock.co/amp/post/164152/flashbots-to-hasten-...
> [flashbots] is not, strictly speaking, a necessary component of the block production process. The centralised Flashbots relay entity does not directly own any hashpower. Nor can Flashbots censor non-MEV related transactions. Miners can still include transactions that did not go through the Flashbots relay.
This is also likely to change as MEV Boost relay code was recently open sourced to accelerate and promote more relay diversity.
https://www.theblock.co/amp/post/164152/flashbots-to-hasten-...
> Therefore, block producers and searchers require complex systems similar to those of a sophisticated trading house, like Alameda Research or Jump for example. This can therefore result in significant centralisation pressures for block producers, with smaller less sophisticated block producers being unable to compete.
I want to make sure I have this straight. The "therefore" in this paragraph refers to how miners "require" a way to front-run orders, bets, and other transactions made on the network lest they fall behind their peers in the mining arms race. The article calls this "Miner Extractable Value" or MEV.
So the centralization pressure from Flashbot comes from miners front-running "DeFi" (decentralized finance) players.
> The Flashbots system is integrated into the current Proof of Work (PoW) mining system on Ethereum. When the Merge happens and PoW mining is switched off, Flashbots adoption will reset to zero again. In order to mitigate against the MEV problem a new Flashbots infrastructure will need to be built up and staking agents will need to adopt this new system. This was not in the ETH 2.0 roadmap and may be an additional complication related to the upgrade to ETH 2.0.
This part is not as clear. How does the MEV landscape change in post-merge Ethereum?
I want to make sure I have this straight. The "therefore" in this paragraph refers to how miners "require" a way to front-run orders, bets, and other transactions made on the network lest they fall behind their peers in the mining arms race. The article calls this "Miner Extractable Value" or MEV.
So the centralization pressure from Flashbot comes from miners front-running "DeFi" (decentralized finance) players.
> The Flashbots system is integrated into the current Proof of Work (PoW) mining system on Ethereum. When the Merge happens and PoW mining is switched off, Flashbots adoption will reset to zero again. In order to mitigate against the MEV problem a new Flashbots infrastructure will need to be built up and staking agents will need to adopt this new system. This was not in the ETH 2.0 roadmap and may be an additional complication related to the upgrade to ETH 2.0.
This part is not as clear. How does the MEV landscape change in post-merge Ethereum?
> How does the MEV landscape change in post-merge Ethereum?
The way that transactions are bundled is no longer done by mining pools and block construction has the ability to be outsourced. I personally, think they should have shipped with this, but I'd expect this to get rolled into a future release.
https://github.com/flashbots/mev-boost
The way that transactions are bundled is no longer done by mining pools and block construction has the ability to be outsourced. I personally, think they should have shipped with this, but I'd expect this to get rolled into a future release.
https://github.com/flashbots/mev-boost
Just to make this clear, the MEV problem and centralization pressure is still there, it's just been temporarily disrupted.
Like you said, this can be fixed more easily now but hasn't yet.
The centralization issue of block formation has been made, temporarily, worse with the switch to PoS [0]. I expect them to clear this up in the future.
Introducing mev-boost now would have likely caused delays in the merge and at some point, you just have to make the switch and fix things later, which is what they did.
[0] https://twitter.com/koeppelmann/status/1570436882483523585
Introducing mev-boost now would have likely caused delays in the merge and at some point, you just have to make the switch and fix things later, which is what they did.
[0] https://twitter.com/koeppelmann/status/1570436882483523585
Thanks, I hadn't seen this distribution.
I have another question.
Why would a centralized relay (flashbots) NOT take advantage of all the orders it processes?
As far as I can tell, the MEV has just moved from the miners to the flashbot(s?).
Why would a centralized relay (flashbots) NOT take advantage of all the orders it processes?
As far as I can tell, the MEV has just moved from the miners to the flashbot(s?).
Reputation?
Folks will stop sending them their TX if they never come out in blocks, but competing copy TX with someone else's address systematically do. So that they do their job is, to a degree, verifiable.
It's a small community who all know each other, so antisocial behaviour can be punished using traditional village politics technology.
Folks will stop sending them their TX if they never come out in blocks, but competing copy TX with someone else's address systematically do. So that they do their job is, to a degree, verifiable.
It's a small community who all know each other, so antisocial behaviour can be punished using traditional village politics technology.
Well, the reputation I now perceive is "abysmal", because everything is served through Cloudflare.
That means Cloudflare can capture or sell the advantage also, unnecessarily.
That means Cloudflare can capture or sell the advantage also, unnecessarily.
The article says it's altruistic.
Not quite. It says this "may be altruistic". "May" does a lot of heavy lifting there.
From an outsider's perspective, a more likely explanation is that the funding is a subsidy. The VC having a stake in this space would have invested in a lot of other companies too. If their funding for Flashbots is notably less than the aggregate expected value of their total investment in the ecosystem at large, and if without Flashbots that expected value would drop by more than the funding amount, then the subsidy would be a net positive investment for that VC.
In other words, they may have seen Flashbots as an ongoing cost of doing business.
And if they had indeed achieved staying centralisation, there's always the access fee structure to hike up.
From an outsider's perspective, a more likely explanation is that the funding is a subsidy. The VC having a stake in this space would have invested in a lot of other companies too. If their funding for Flashbots is notably less than the aggregate expected value of their total investment in the ecosystem at large, and if without Flashbots that expected value would drop by more than the funding amount, then the subsidy would be a net positive investment for that VC.
In other words, they may have seen Flashbots as an ongoing cost of doing business.
And if they had indeed achieved staying centralisation, there's always the access fee structure to hike up.
Right now, because there's no other way to cash out. The ETH you get for your stake stays locked up with your stake!
The only ETH a validator can get from validating that they can sell immediately is their take from MEV, i.e. frontrunning.
The only ETH a validator can get from validating that they can sell immediately is their take from MEV, i.e. frontrunning.
Incorrect - since the merge validators are been getting ETH1 transaction fees & block rewards (even non-mev validators)
as of today, only 21% ( 90537 of 429393) of validators are connected to one single server.
source: the single server https://boost-relay.flashbots.net/
We are in mev-boost infancy and there are already 8 relay to chose from, beside home-building of blocks.
source: the single server https://boost-relay.flashbots.net/
We are in mev-boost infancy and there are already 8 relay to chose from, beside home-building of blocks.
ITT: People see a crypto-based headline and write smarmy, reddit-tier, bad-faith comments about something they know nothing about, and don't desire to know anything about.
I can see it's already starting.
I can see it's already starting.
This is a deeply misleading headline. Other people have explained it as well, but Flashbots is a relay for transmitting transactions to the Ethereum network in a way that preserves their privacy (and exposure to something called MEV[1]). It's purely optional and the vast majority of transactions[2] do not actually pass through this relay. MEV is an interesting area of research[3].
Unrelated aside: why is the quality of Hacker News posts and comments on crypto so bad? I experience extreme Gell-Mann amnesia[4] almost every post I read. This is the industry I work in and the amount of misinformation produced and spread on Hacker News is absurd.
[1]: https://arxiv.org/abs/1904.05234
[2]: https://explore.flashbots.net
[3]: https://flashbots.notion.site/flashbots/mev-day-836f88806995...
[4]: https://www.johndcook.com/blog/2021/01/18/gell-mann-amnesia/
Unrelated aside: why is the quality of Hacker News posts and comments on crypto so bad? I experience extreme Gell-Mann amnesia[4] almost every post I read. This is the industry I work in and the amount of misinformation produced and spread on Hacker News is absurd.
[1]: https://arxiv.org/abs/1904.05234
[2]: https://explore.flashbots.net
[3]: https://flashbots.notion.site/flashbots/mev-day-836f88806995...
[4]: https://www.johndcook.com/blog/2021/01/18/gell-mann-amnesia/
> Unrelated aside: why is the quality of Hacker News posts and comments on crypto so bad? I experience extreme Gell-Mann amnesia[4] almost every post I read.
It's because the posters think they are smart. With every passing day that cryptocurrencies become adopted/accepted, their smartness becomes questionable, even to themselves. For them, it is easier to double down on the flawed arguments - well over 10 years now - than accept that while they are smart in some areas, they are not in all.
It's because the posters think they are smart. With every passing day that cryptocurrencies become adopted/accepted, their smartness becomes questionable, even to themselves. For them, it is easier to double down on the flawed arguments - well over 10 years now - than accept that while they are smart in some areas, they are not in all.
Afaik, just because miners choose to accept flashbots bundles doesn't mean they don't also accept mempool transactions. This isn't mutually exclusive, is it?
Hm. Have the economics changed? In a proof of work system, you have to operate a mining farm, using hardware and power. In the new proof of stake system, you just have to operate a validator, which is cheap, and own lots of ETH. So, while there are advantages to front-running, there's not as much cost pressure to do it. The motivations of miners and HODLers are different.
>there's not as much cost pressure to do it
This is false. You never had to run a validator to watch the mempool and do MEV.
This is false. You never had to run a validator to watch the mempool and do MEV.
And that server was: Cloudflare:
https://relay.flashbots.net/
https://relay.flashbots.net/
clickbait title. the network doesn’t need flashbots to function, it can produce blocks if it goes down.
Whatever happened to the decentralization concept of every user being a node? Like not just computers but apps for phones too, so in order to use it you also have to share some resources in return. Is BitTorrent really the closest thing to decentralization?
There is still a plan for that, but it has required a lot of R&D and 'invention' of new techniques. Read up on Verkle trees [0]. That said, the plan is to basically prune the database and not keep history of what each wallet balance was at a certain block. This storage is going to be pushed off to indexers that want to provide it as a service. I'm not sure that is a good or bad thing, but it'll enable the future you're talking about since we do not have infinite storage/bandwidth on our small devices.
[0] https://vitalik.ca/general/2021/06/18/verkle.html
[0] https://vitalik.ca/general/2021/06/18/verkle.html
The moment you have to stake 32 ETH to be a validator, it was already made highly inaccessible for most.
As soon as that node ends up running on AWS rather than a user’s typical laptop as suggested by Vitalik and his friends, then that was another decentralization failure there and there is no self-hosting at home.
Finally, when the merge was finalized, there where only two addresses that ended up holding more than 48% of Ethereum that is used for the PoS mechanism to work. Another centralization risk feared that happened in reality.
So, that looks to me like a failure of decentralization and a more centralized system that as expected; made worse with Proof of Stake.
As soon as that node ends up running on AWS rather than a user’s typical laptop as suggested by Vitalik and his friends, then that was another decentralization failure there and there is no self-hosting at home.
Finally, when the merge was finalized, there where only two addresses that ended up holding more than 48% of Ethereum that is used for the PoS mechanism to work. Another centralization risk feared that happened in reality.
So, that looks to me like a failure of decentralization and a more centralized system that as expected; made worse with Proof of Stake.
Now Irrelevant
Somewhat counterintuitively, PoW is inherently centralized. Its properties that make it viable at all also make it easy to become controlled by small number of big actors. This is hardly specific to Ethereum.
Single point of failure
It's actually not a single point of failure, there's a graceful fallback where the client builds a block locally.
DanAtC(1)
So lets talk it straight.
The merge killed the thing that keep ETH DE-centralized.
Now who owns more ETH has more power...
Like centralized finance.. instead of Defi.
The merge killed the thing that keep ETH DE-centralized.
Now who owns more ETH has more power...
Like centralized finance.. instead of Defi.
In Ethereum, as of May 2022, it is believed that over 90% of miners are connected to the Flashbots server. One can determine this by observing that over 90% of blocks contain Flashbot transaction bundles. Adoption from 0% to around 90% occurred almost entirely within 2021.
Flashbots is a single centralised entity. It is funded by Fred Ehrsam’s crypto VC firm Paradigm, although it appears that this funding could be altruistic. Searchers send there MEV transactions to the server with the address https://relay.flashbots.net/ and the server is currently located in Ohio in the United States of America.
Flashbots is a single centralised entity. It is funded by Fred Ehrsam’s crypto VC firm Paradigm, although it appears that this funding could be altruistic. Searchers send there MEV transactions to the server with the address https://relay.flashbots.net/ and the server is currently located in Ohio in the United States of America.
flashbots faq seems out of date as relay.flashbots.net resolves to 104.26.10.65 & isp = Cloudflare in San Francisco
On port 80 that makes sense, but is the actual relay on port 80? Or is it a web API? If the latter then I am not surprised they chose to put CloudFlare on top of it.
Domains resolve to the same IP disregarding port number.
Unless you have the really expensive TCP proxy from CloudFlare it is typically only used on web protocols.
Yeah, I kept thinking about it after writing my comment ...
"Well, but maybe if you ..."
But I've never seen it in the wild.
"Well, but maybe if you ..."
But I've never seen it in the wild.
I see a JSON error straight on HTTPS. Must be an API directly.
But HTTPS is 443, not 80.
The fact that Cloudflare was privy to all Ethereum ecosystem communication with flashbots gives me the heebie jeebies.
But HTTPS is 443, not 80.
The fact that Cloudflare was privy to all Ethereum ecosystem communication with flashbots gives me the heebie jeebies.
Yeah had a bit of a brainfart there about port I really just meant presence on the web…
The backend could still be in Ohio. Just fronted by CF. Which is centralized.
Mining pools, not miners, connected to the flashbots mempool to pick up their transactions. There can be many mempools and mempools don't need to talk to each other. It was in the benefit for the miners for pools to do this (ie: their customers), so all the big ones did it.
By doing so, miners would earn extra value (aka: MEV) from these transactions because flashbots would bundle them up and order them. If you wanted to be at the top of the list with your transactions, you'd pay a bit more.
The really centralized bit of this is that flashbots actually can be turned on/off [0], which I find rather inexcusable.
[0] https://twitter.com/TimBeiko/status/1570435474535354368