I think he's misconstruing what Sergey is saying about market coverage and not providing the full context on oracle security and the Uniswap TWAP
Thread for context below 👇
2/ Sergey was emphasizing the importance of proper market coverage
This does not mean taking a simple median or mean across exchanges
But instead aggregating data from all trading environments by taking into account liqudity, volume, time and other differences across exchanges
3/ Market coverage is important because it ensures price data reflects the true global market-wide price and not just that of one or a small number of exchanges
Chainlink covers this topic of market coverage and data quality extensively in this blog post blog.chain.link/the-importance…
4/ Hayden seems to be misconstruing what Sergey is saying
It's true there is natural exchange concentration and it's exactly why we need proper market coverage
Single source price feeds may provide coverage initially, but there is no guarantee it will be maintained over time
5/ If done correctly, tracking more exchanges will only ever increase the manipulation resistance of the price oracle, not lower it like Hayden suggests
Below examples are what I believe Hayden is talking about, but this is not how Chainlink achieves market coverage
6/ Chainlink price feeds are able to achieve full market coverage because data is fetched from numerous professional data aggregators who track all venues and have full time monitoring teams with 24/7 alerts
This smooths outliers and ensures the true market wide price is used
7/ Without proper market coverage, you lower the cost of manipulation dramatically and allow for inaccurate prices to be published
This is exactly what happened with Compound, which only tracked a single exchange Coinbase and was manipulated because of it chainlinkgod.medium.com/proposal-for-c…
8/ I'm emphasizing this point because I think it's important to understand why market coverage is so important
Price feeds that lack market coverage have already been exploited time and time again, with now over $100M in user loses in DeFi decrypt.co/49867/defi-fla…
9/ Market coverage is important regardless of where the data is sourced
While it's true there are different attack vectors, you can absolutely compare the differences between a price oracle with market coverage to one without the proper coverage
10/ Now this is where I believe Hayden is being a bit misleading to readers
The Uniswap Time Weighted Average Price (TWAP) oracle only provides data that is accurate according to the Uniswap market, and NOT the market wide price which protocols need
An important distinction
11/ Uniswap TWAP oracles are better than the Uniswap spot price (no more flash loan issues)
However, it still only tracks data from a single exchange and does not provide adequate market coverage, which as I described above lowers the cost of manipulating the price oracle
12/ Providing data of a single exchange isn't useful if that exchange is out of sync with the rest of the market
It's the same "garbage in; garbage out" problem that market coverage solves
Pulling data from just a single exchange always ends up with lower quality data
13/ Also of note, TWAP oracles are not a replacement for the VWAP or Spot price
TWAP is a lagging indicator that becomes inaccurate during times of high volatility and requires a fundamental trade-off of tamper-resistance versus freshness, an undesirable property for users
14/ With the increase of MEV, Uniswap TWAP is not as highly tamper-proof against manipulation as stated
With enough value at stake, it can become profitable for miners to manipulate the ordering of transactions and inter blocks to manipulate a TWAP snapshot
15/ The mixture of low market coverage and the inability to scale price oracle security without sacrificing accuracy makes the Uniswap TWAP oracle a niche offering
It is not the general purpose price oracle people think it is and not a replacement for Chainlink price feeds
16/ Chainlink is secured through its cryptoeconomic security where malicious oracles experience a large opportunity cost in the form of lost future income
Oracles are also public entities with their reputation on the line, which is fundamental to their business model
17/ Chainlink oracle nodes are operated by data providers and professional DevOps teams that rely upon their good reputation for revenue
Being malicious means they forfeit their entire business as they would lose all user trust as well as their delegators/stakers/clients/users
18/ Additionally, Chainlink's cryptoeconomic security is why its price feeds are able to operate even during extreme network congestion, the profitability provides both a high opportunity cost for maliciousness as well an incentive to provide the most reliable services possible
19/ This point is incredibly misleading as Chainlink price feeds combine data from a multitude of data aggregators, who each track hundreds of exchanges (both CEXs + DEXs)
Chainlink is decentralized at the node operator & data source levels to prevent any single source of truth
20/ Decentralization is achieved when you combine multiple independent components and becomes no longer centralized, which is what Chainlink does
Additionally, Market coverage means you need to aggregate from ALL types of exchanges and not just one particular ones you like
21/ I love DEXs, especially Uniswap, but the reality is that CEXs and custodial exchanges are not going away any time soon
Proper market coverage doesn't care about your prefered form of trading, but needs to track all environments that users choose to deploy their liqudity into
22/ Again, this is missing the importance of proper market coverage
AMMs will improve no doubt, but that doesn't change the fact that price oracles that pull data from a single exchange are cheaper/easier to manipulate than those that aggregate from all exchanges
23/ A specific AMM may be more expensive to manipulate than a specific CEX, but will still inherently be cheaper to manipulate than the global market-wide price
With volume shifts across exchanges, pulling from a single exchange is simply reckless for high value contracts
24/ I don't think the ad hominem attacks were needed here, we're all adults here
Hayden has already blocked me for simply disagreeing with him on this topic of oracles, which is unfortunate, but I will continue to educate about the risks of single source price oracles
25/ I hope this thread provided you some important context into the naunced topic of oracles
While I still deeply and fundamentally disagree with Hayden and his approach to oracles, I still wish him all the best and am interested to see what Uniswap v3 will bring
Cheers frens
• • •
Missing some Tweet in this thread? You can try to
force a refresh
I originally attempted to post this proposal on the Compound Governance Forum, however after over 22 hours, it is still "awaiting approval" from the Compound team
As a project based around “community driven governance,” I find it concerning I am being censored in this manner
For the sake of transparency and due the urgent nature of the matter, I am posting this proposal on my own Medium page so the $COMP community can decide for themselves if this is the path they want to take.
1/ "The goal is to transition to an active piece of IT infrastructure, empowering the document to communicate with external resources such as data feeds, payments infrastructure, ERP, and CRM systems" -@tgonser, founder of @DocuSign
3/ Tom understands the value of smart contracts and the importance of secure oracles to connect these contracts to real world data, events, and payments
Take a listen for yourself
In fact, DocuSign has been interested in smart contracts since 2015
1/ Creating Definitive Truth, a thread giving insight on Sergey's recent presentation at #SmartCon 2020 covering how @Chainlink generates definitive truth about the real world, the acquisition of deco.works from @Cornell, and much more
2/ Chainlink oracle networks operate at the speed and latency they do because of the limitations of Ethereum L1, not because of any limitations of Chainlink
Ethereum L1 is simply where all the demand for external data currently is
I've just exited all my yield farms, the risk adjusted returns just aren't there anymore
For good yield you now have to deposit into pool2 (lol no thanks), be the first to hear about a new crop and ape in (rugpull risk), or stick with pool1 for a measly 100% APY (not worth IL)
It was inevitable that these foodcoin ponzis would come to an end, however if you know how to play these games right (as I've taught), you should be well in the green by now anyways
I'll still be on the lookout for new crops, but as things die down, lower your risk, not raise it
As time goes on and yield continues to drop, the rugpull risk will continue to rise, lowering risk adjusted returns even further
Find out who the suckers are, because if you don't, you're probably the sucker who will get left holding the bags