The Shape of Protocols to Come by Tim Beiko | Devcon SEA

[Music] okay oh nice um thank you for making it to the real ethereum 3.0 talk um hi I'm Tim moo I work at the ethereum foundation on core protocol coordination and as part of this um I spent a lot of time thinking about how we should Steward ethereum like what are the things we need to do understand to make sure that the protocol evolves in the right direction and a few years ago I was quite frustrated because I felt like we didn't even have the right mental models or the right Frameworks to think about ethereum and the reason for this is like ethereum is quite hard to describe right like uh it's this complex thing and so it leads us to use analogies to describe it so to use one myself um oh nice I have sweet to use an analogy myself um I really like the parable of blind man of the elephant to think about this so when different people look at ethereum they'll see something different right some people will look at it they'll see that there's this asset in the middle ether and think of it as money um other people will see that oh there's like this evm it runs computations it's like this world computer um you can think of it as a settlement layer for everything to use as sort of This Global Ledger um and even extending that something like ethereum is a digital nation state and all of these things are partially true but there's there's ways in which they sort of fall apart if you look at them a bit more closely um obviously ether is valuable it's like an asset you can buy and sell it but it functions very different than say Fiat currencies similarly computation runs on ethereum it's distributed but the intuitions we have about scaling ethereum are very different than how you'd approach scaling like a normal computer system um and then when people start talking about ethereum being a country or anything like that you know ethereum doesn't have a border doesn't have citizens doesn't have like natural resources to protect and so if we don't have actually good mental models to think about ethereum we risk sort of misunderstanding what the thing is and sort of doing a poor job of shaping it in the future so couple years ago I set out to find better ways to go about describing this ethereum elephant and one intuition we had around this is that maybe we should zoom out a little bit and instead of looking into like every part of ethereum and trying to understand how it works could we find something a bit broader and effectively find like a higher level class which in theorum would be just a single instance one inspiration we had for this was the idea of like chaos theory if you look at Chaos Theory you have these abstract formulas that allow you to describe systems ranging from like biological organisms to financial markets or another example could be something like economics people obviously bought and sold things before we had a formal field of Economics but once we had the concepts like say Supply and Dem demand and how markets worked it allowed us to reason about what was happening in a much clearer way and so in this case the framework we landed on was protocols after some initial discussions a few of us within the EF and outside had confidence that like protocol might be a rich enough topic that if we studied it as like a first class thing we could derive some general insights that we could then apply back to ethereum helping us to guide its stewardship and so over the past two years we effectively were inspired by this idea of uh de blindman and the elephant and decided to fund over 50 researchers to study protocols from many different angles and try to give us reports around what they were seeing so that we could see is there actually a common thing there we people from with backgrounds in architecture law engineering uh Environmental Management arts and way more and they each ort of spend time collecting these findings and you can think of this as like field notes from touching the elephant in different places and now I'd like to share some of their findings you can think of this as like a sketch of the shape of protocols so first off let's look at how protocols get adopted protocols tend to have like a pretty slow adoption curve especially early on and the reason for this is that the tend to require consensus across many different people in different institutions so this means that at first you can feel like it's it's it's almost like stagnant like there's not a lot of energy there's not a lot of process but then if the conditions are right you can have very rapid adoptions of protocols everyone has like this common knowledge about things and so it can involve very much like a phase shift and because you have this common knowledge once you've actually adopted the protocols it feels like it can get very entrenched actually so it's like a new status quo in a way so one way to sort of visualize this is think of it as like a reverse Z curve where early on you have like this initial Wiggles of adoptions and then um this sort of phe shift happens where the protocol gets adopted and then it's a new equilibrium and the reason for this is just again because there's like all this common knowledge that's built up across people around why uh you know this protocol should be adopted and they've coordinated on it so one concrete example here is um how time zones got adopted for decades there were conversations around time zones as uh as uh there were more and more trains in rail worlds across the US specifically in Europe but it didn't really go anywhere until one year there was actually a conference which was sort of the Tipping Point where at that point in time people decided okay we're going to do time zones within a year after that 85% of the US had time zones adopted so it sort of went through this phases shift within the span of a year and now if you think of time zones it's effectively this protocol we take for granted and even doing major changes like trying to change daylight savings time is this very complicated thing it's super hard to move from the equilibrium and one implication of this is that once these protocols get adopted they'll tend to fade into the background as researchers were studying protocols there was this one quote that came up over and over uh different domains um that everyone would sort of use as like an anchor for how to think about protocols it's a quote by a an Whitehead that reads that Civilization advances by extending the number of important operation they can perform without thinking of them and so we started thinking of protocols as effectively these tools to generate these Whitehead advances to allow civilization to perform more complex operations by abstracting them in the background one example of this is like GPS is this very complex technological protocol um that we've deployed across the world and now we sort of take for granted that you can just you know position yourself between two points in space and not really have to think about it and to use an example a bit closer to us you can think of UN swap is another uh as another one where prior to Unis swap you know deploying markets between assets was this long complicated process whereas now we have like a simple formula that we've protocolized and you sort of take it for granted that when a new coin appears there's like a liquidity pool that's Pro that's powered by the unisol protocol um and one implication of this is that almost like the better a protocol is the less visible it becomes over time think of something like HTTP you can argue HTTP is like the greatest digital protocol that's ever been invented it Powers effectively the entire internet but when I talk about it I don't get like a strong emotional reaction similar to how like when you use a great product you know you notice it or if you see like a beautiful work of art or you walk into like a nice space like the conference Vue um HTTP sort of Falls flat and you're like yeah it's a good protocol but on the other hand think of bad protocols uh one obvious example is like air travel protocols if you're traveling you know internationally and I put these pictures here you immediately have like a gut reaction of what it feels like right to take off your shoes you're scared to be pulled aside um and maybe you know this goes well you fly but then you're trying to get a taxi to go back to your airport after your jet lag and if the protocol there is bad you know you immediately feel it right you're studing there outside uh it's warm and there's 100 people waiting for their grabs um and so this was a really valuable Insight because um it allows us to sort of or it sort of Illustrated a different lens through which we could start viewing protocols one one of the researchers formalized this and called this the CFA index the idea being that it's much easier to reason about a protocol by imagining its worst version across many different dimensions than to try and imagine the best possible version and again this is very different in your intuition when say Building Products if you're building your product you're trying to think like okay how do I build the best version on across all these Dimensions you know I go from like zero to 10 Protocols are a bit like the opposite it's more like you have all these dimensions and you're trying to get from minus 10 to zero and when I was thinking about this this week um it hit me that this is kind of why the Bitcoin protocol I think is is so elegant if you look at the Bitcoin white paper um the second line or so uh focuses on what is like the core thing it's trying to address um so soshi's claim is you know trusted third party are like the worst things about online payment system and the entire protocol is focused on just trying to get around that one thing if you look at the bottom of the abstracts it sort of says look there's all these other parts of the protocol and they sort of kind of work um but it's sort of less important than the idea that on this one bad dimmension we're trying to optimize to uh go from minus 10 to zero and flowing from this like if we're trying to like protect these protocols or get them to avoid these bad States another way that's useful to think about them is to to treat them as Commons in a way Commons are these shared resources that have a risk of like being captured or overused um in in sort of outside of crypto people will use this when talking about things like Fisheries or P or pastures basically natural resources um and obviously in our context uh these protocols tend to not have these like physical boundaries um there's things that have clear digital boundaries you can think of ethereum itself as having like clear digital artifacts or even if you consider Linux to be a protocol you know it's like a clear thing to point to but there's also some more diffus things for example think about like the evolution of workplace safety protocols it's clear that there's like a common good here you know there's like this sense of safety and injury and whatnot but it's distributed across a whole range of Institutions um without necessarily having like a single artifact but still I think it it points at an important intuition that like if you want to avoid the bad cases just like for physical Commons we need to have strong stewardship because we want them you know we want to protect against capture overuse and other risks and the sort of risk a capture let us to one of the sort of final insights around what's unique about protocols at the very start of this we spent a lot of time trying to Define what even was a protocol and especially how it was different than other things you know like what's the difference between between a protocol and an API or a standard or a grammar and one thing that became clear through this work is that protocols have a notion of conflict embedded sort of at their core um and you know you can see this very quickly in something like ethereum right ethereum is effectively this digital protocol we have nodes on the network which have to come to consensus about the state of things so they have all these ways to resolve conflict you know by peering your or un peering from each other there's some more um explicit parameters they get to NE negotiate over like the block gas limit um but beyond that we also have these like Mela protocols around ethereum that exist the mediate conflict like how we plan upgrades and again outside of crypto you could think of something like traffic rules right the way that traffic is organized and we're trying to mediate different people going to different places but going through the same roads um is another way that like we have these protocols that manage conflict and this was like such a central point that it allowed us to sort of simplify our working definition of protocols about an order of magnitude so in 2022 when we started this Danny Ryan had the following definition for a protocol which I'll have to read um which is astrum of cified behavior that allows for the construction of emergent or emergence of complex coordinated behaviors at a jent loai and so this was broadly right but it was quite verbose and then through all of this work at the end of this summer vat who runs the summer protocols program was able to summarize it in two words which is Protocols are engineered arguments and this is like a really helpful working definition of a protocol especially when contrasting yet with things like standards apis and so on and so to recap what we had so far you effectively have like these five initial properties of what the shape of protocol starts to look like we have this slow adoption and entrench persistence that we first covered sort of reverse Z curve then the idea that protocols create these Whitehead advances that they provide sort of this invisible infrastructure that Civilization can then build on top of um we have this idea of a CFA index where you can analyze a protocol by looking through all of the worst versions of it across different dimensions then thinking of protocols as Commons which have a risk of being captured and therefore require strong stewardship and lastly treating protocols as engineered arguments so realizing that they embed conflict in many different ways and this sort of validated the initial thesis that there was a common thing around protocol that applied across different domains that we could use to discuss um to discuss our problems within ethereum and hopefully learn from others as well and so now I'd like to talk a bit about how protocols May evolve in the future and specifically how ethereum may help shape that and to do this it's useful to like zoom into ethereum's distinguishing property as a protocol uh which we which I think is hardness this is a term that Josh dark coined a couple years ago um and defined it as the capability to make the future more certain ethereum isn't the only source of hardness in the world and in this piece uh Josh puts out three categories so atoms institutions and blockchains and but ethereum's hardness is special in three different ways first is that it's globally homogeneous meaning that no matter where you are in the world you have like the same degree of hardness when you use ethereum and this is very much not true of atoms and institutions the second is that it's independently auditable so anyone that uses ethereum for hardness is able to verify whether or not a certain thing happened and then lastly it's permissionless accessible so you don't need sort of a high barrier to entry to use the the and so if we use this and understand like ethereum has hardness as its core um we we can start to see how it can shape the future of protocols the first is as an hardened foundation for other things um this is kind of obvious for anyone who's at Devcon here but it's striking how special and unique this is there's not many other things in the world that exist to provide sort of the strong source of persistence and reliability to financial assets cultural artifacts or even a whole set of protocols built on top of it so um again this it's kind of hard to overstate this but like looking from all these different domains it feels like something truly unique that we should not take for granted but stemming from that there's like a more subtle way in which uh we we we have sort of hardness at our core and it's as a hardened culture and here you can look at hardness both in the way I just defined it as like uh giving you better certainty about the future but also in it's like more traditional sense as battle tested ethereum operates in a pretty adversarial environment and over time we've developed like norms and practices to mitigate and respond to attacks at both the protocol application and infrastructured a but in parallel to this we also have a culture of looking really far into the future of trying to like plant seeds for the ecosystem to grow to become resilient and not dependent on a single uh institution or set of people and also have a technical road map that's sound in the long term and so by combining both like these hard Foundation as a protocol and this softare like cultural hardness um ethereum effectively can impact the future protocols Beyond just crypto and one way one way I thought about this is like ethereum is effectively this tool that we can use to harden Commons um outside of the space as well and this is because we have this opin platform that's very large and and and sort of very versatile but that's also fairly sandboxed from the rest of the world this means we can run experiments that would be probably too risky to happen anywhere else but it can still happen at sufficient scale on ethereum and provide value to the rest of the world two examples here on this slide are quadratic quadratic voting and prediction markets both of these started off as these theoretical ideas that didn't quite have a place to sort of be prototyped in the world eventually e theorem became that place we were able to run small experiments and learn from those improving our understanding and then you know leading to much more significant outcomes most notably uh you know prediction markets playing a major role in a recent us pres election um and so I think going through all this and coming back to the original question I feel like we have like a start of an answer to how we can best Steward ethereum so by understanding um the the notes are wrong on this slide but we'll run with it um but okay but understanding anyways the shape of protocols um we can better calibrate effectively how we approach developing ethereum in the future um you know we should expect that things will take time especially early on but we shouldn't be too surprised if things gradually happen as a sort of phase shift all at once we should expect conflict and capture and engineer systems that gracefully handle them and instead of building for hypothetical Utopias we should be honest about everything and try to solve for failure modes in the protocol and by having this shared language this shared framework of of protocols we're able to learn from other communities who otherwise wouldn't be interested in our problems and at our best we can also export our best practices out to them so in other words the understanding the shape of protocols is the way ethereum community can leverage its impact Beyond crypto thank you thank you very much Tim fantastic you win the record for the first speaker that has finished Before Time wow every other speaker pushed me to come on stage a bit earlier than expected but nevertheless that gives us a little bit more time for some great session of Q&A I was going through the questions some questions are not exactly linked to the topic so if possible we want to keep the topic and uh as our Focus so please ask questions regarding what Tim has shared about protocols ethereum um and feel free to upvote as well right if you see a question that looks really good please upvote it so that our speaker can address it let's go with the first one on top how do we convert merge on a conception of protocols which will be sufficiently specific to lead to useful actionable in oh went actionable oh my they keep okay hold on uh I can't scroll down can I scroll down I don't know if the team can scroll down cuz I yeah yeah the third one okay actionable insights while keeping it well keeping okay I think I got the question I'm happy to answer it okay let's go to number six do you think e ecosystem operation is a bad protocol because we don't see any feedback Channel that's a good question Tim how do you think you'd like to address it yeah I I guess I dispute the premise I don't think that like there's no feedback Channel I know I work at the EF and I read all the tweets and I listen to everything so I think um a lot of it is not necessarily visible you know the EF doesn't have a suggestion box but we get a lot of unsolicited feedback so we don't necessarily need to um and and then I think back to what I said earlier like um and you know back to is talk this morning um getting the feedback doesn't necessarily mean you have to act on it right like I think one thing the EF wants to optimize for is um leaving the space for others to grow and and sort of take ownership about Solutions in the space and so if we just reacted proactively and solved every single thing um others would sort of feel like they don't have the space to do that um but yeah to the extent to which we have or or we don't have good feedback channels then yeah that would make it a bad protocol that's a good answer and I want to pick up on that because our previous session with Justin a very big crowd came in and he was sharing that if you want to accelerate growth come and email us and join the team don't just watch from the sidelines you know giving feedback is good but participating I think even better great question great answer let's look at the next one um okay what are the takeaways for ethereum here rules of Thum for healthy protocol Evolution past mistakes we should avoid repeating yeah um one thing I felt good about doing this research is honestly a lot of the in instincts of the ethereum community are are pretty good like so um it's it's it's it's made me a bit less concerned about this but um you know like probably the biggest one is like this idea like solving for the negative versions and um this is maybe not as visible to to like people who don't work on a protocol every day um but the main thing that we talk about when we discuss uh protocol changes for ethereum is security right like everyone comes in with a new idea for a new feature for ethereum um it's it's always great it provides a ton of value to user but there's a security issue with it and effectively not compromising on ethereum security is more important in the protocol development process than trying to deliver as much feature or as much value as possible through features um so that's one of the takeaways we like it's kind of nice that we were already sort of doing it um I do think the adoption bit is something we should think a bit more about where a lot of the intuitions we have around how ethereum will get adopted we probably borrow from like the technology sector and product development um and we should probably be looking more at things like yeah how time zones got adopted or how GPS got adopted than uh products y excellent answer all right number four votes on the top how does engineered argument differ from logic defined by wiky as the study of correct reasoning um yeah I'd say it's like correct is uh correct is subjective right like you know uh and and one of the big things is just this idea that protocols exist to effectively mediate conflict um so everyone thinks their version of the protocol or their view in the protocol is correct but a good protocol will be able to uh to deal with cases where yeah it's incorrect all right thank you okay we're back to our earlier question which I couldn't finish in time hopefully now it'll stay on the board don't vote for anything else wait hold on how do we Converge on a conception of protocols which will be sufficiently specific to lead to useful or actionable insights while keeping it open enough to facilitate input of offchain perspectives well um how do we Converge on it you got 30 so I this is like a low confidence answer but I think at this point it feels less like there's a single core formula that we can use that will be like perfectly actionable rather than like these many different lenses so I think um one thing that's been useful in this summer protocols research is having people from different domains to be able to validate like is this an ethereum thing or is this like a broader phenomena and yeah I I think just like gut checking against other domains and yeah and I think maybe just one more question the second one on the list uh in case anyone's interested where was the research published in one place and where was the research published summer protocol.com all of the research is there um yeah if there's one place to go that would be it fantas