Almost a year after the SFBW talk outlining our ideas on fairness, this is now getting real. I’ve made some updates to my research around Fairness as well as a prototype implementation, and will be publishing Wendy Grows Up next week (version 2 to Wendy, The Good Little Fairness Widget). I wanted to share the white paper with the Vega community first and would love any feedback you have. You can download it here. Fairness has become a hot issue in the last months – especially on Ethereum – due to increase attention of Miner Extractable Value (MEV), i.e., miners profiting from reordering or censoring transactions, as well as sniping bots scanning unprocessed transactions for potential opportunities and then outspending the originator to preempt them on the transaction. With increasing volume of DeFi – both in terms of value and number of products – these issues get even more important, and might become a roadblock.
*Wendy Grows Up addresses this problem in a flexible way that can easily be adapted to different blockchains and application needs; it can serve as an add-on or an oracle for chains like Ethereum, or be integrated with low overhead into high speed chains like HotStuff or Tendermint.To further refine it I am looking for feedback from traders and liquidity providers about what parameters you are looking for in fairness. Please drop me a note with any thoughts you have. I will be presenting Wendy at the ACM conference on Advances in Financial Technologies on Wednesday 21st October 2020. If you are interested you can register here We are also holding an AMA on Wendy on Tuesday 27th October 5pm BST … it would be great to see you there.
Thank you for sharing this with us. This is such a fascinating area of research and in my opinion so important. The crypto traders learnt to live with many “quirks” of transactions processing and power that miners are holding over the execution of trades. Many traders/investors I know were frustrated over this issue countless times. It’s amazing that Vega is looking to improve this.
I only have one minor question: are there some limitations to this system that might occur before the validator sees the transaction for the first time? The system kicks in once the validator notices a transaction and assigns a sequence number to it. In scenarios where companies buy server space close to the exchange, their transactions will be executed quicker. In a decentralised setting are we assuming that the number of validators is high enough to make that system fair (in terms of assigning a sequence number to transactions)?
thanks for your question. With the current design, having a premium connection to one validator isn’t buying that trader much, as that doesn’t give them any advantage for other validators (in fact, we’re assuming that some traders can
run a validator and thus have negative transmission time). Being close to all of them is hard, because then your own nodes need to coordinate (which costs time). So far, I see three strategies a high speed trader can still do to get ahead anyway:
Figure out which part of the planet is the best to be to have the lowest average message transition time to the validators, and then sit on a very, very fast connection
Buy a telco and mess with the Border Gateway Protocol; that would essentially be our asynchronous adversary who can manipulate message delivery times all over.
Have a node next to all validators and (deterministically) generate the same transactions everywhere.
As some speed advantage is always possible, we’re currently thinking on ways to negate too small advantages; if, say, we add a random number of 1-10 milliseconds to all transaction timestamps, there is no business model for spending millions to get 1 tenth of a millisecond ahead (there was
a talk on AFT 2019 describing some ideas to this end for centralised exchanges).
Another idea we’re toying with is to allow discrimination based on stake - in this idea, if I havea market handling corn prices in Australia, an actual Australian corn farmer would always get precedence over a day trader in New York. Which will lead to fun research proving online that you’re a corn farmer in Australia.
In the end, this needs to be decided by the marketmakers though - if a lot of marketmakers say they’d like such a fairness concept for their markets it is easy to build the option into Wendy; if they say they want the opposite and let traders pay to be preferred (in which case the fee at least goes into
the ecosystem and not into sellers of fibre lines), that also is an option that can be integrated into Wendy.
