ZK7: ZPrize and Accelerating Zero Knowledge - Alex Pruden – ZPrize

Transcript

[Music] excited to introduce alex who's going to be telling you something about z prize so please take it [Music] away hey everybody i hope everyone had a good lunch um my talk is only 15 minutes so not enough time for you to fall asleep thankfully so uh no hey thank you very much everybody for being here so this talk is about the z prize initiative that i'm helping to organize um the whole purpose of it is to accelerate the future uh accelerating the future for zero knowledge cryptography but before i go into the talk i just wanna i want everybody in this room to just take a minute and consider how quickly and how fast and how far the

space has come in just the last four years in large part due to the efforts of the people sitting here and the people downstairs and the people this conference so just give yourselves a hand just real quickly you know this i mean just in the short time that i've been following the space it's gone from being you know kind of a you know something that was of academic interest is something that's being practically applied um you know throughout throughout the industry and so it's fantastic to see now that said um i think we can all acknowledge that we still have a way to go to scale this technology to make it practical for everyday users at the scale of web 2. so and that's really what the z prize is all about so that's what i'm going to be talking to you guys about today so

what's the problem so making zero knowledge cryptography practical what does that mean so it's you know it's kind of you run into the problem of well what's the application for you know the the zk crypto that you're using so there's different applications different applications potentially have different requirements different companies maybe have different attitudes towards like what they care about optimizing different proof systems like you know some of the some of which have been kind of talked about earlier today have sort of different unique features so what you know it's thinking about how to accelerate with zero knowledge cryptography kind of it's a little bit too broad of a bucket um and so that that's sort of the problem that we were considering as we started thinking about z prize so first one question is how do we define success what does it mean to have a faster

implementation of marlin or something you know or you know what does it mean to accelerate zero knowledge cryptography and so i think there's there's a couple different ways to look at it one is minimizing latency you know so what's the shortest amount of time that you can produce or verify a proof in the other is maximizing throughput right how many proofs can you generate in a given time frame or verify and then the last is maximizing efficiency right and different applications potentially require different optimization objectives right for example if you're doing a rollup um maybe it's like throughput for verification is much more important than latency right whereas if you're doing something like a leo proof of succinct work mining the latency to produce a given proof specifically for a coinbase transaction becomes the more important thing so how do we capture all

of you know all of the things that people might care about kind of under one umbrella of a competition to uh you know to make to to achieve these goals and then the other thing is to make it even more complicated you know how are we measuring the success we're talking about accelerating zero knowledge cryptography you know again like what are we talking about and there's again different levels at which you can think about solving this problem right there's first and work that many people here have participated in is designing new proof systems that are more efficient that use primitives kind of more effectively to achieve some outcome there's also the way it's implemented right someone was joking earlier about the number of implementations of planck there are that are out there um you know there's obviously not just the implementation of a given proof system but even the lower

level primitives that compose the proof system um you know there's a huge uh variety of things there and there's potentially ways to uh to optimize um and then lastly is the hardware so i think for many of us in this room and including some folks who are in the industry working on this problem specifically it's really exciting to consider um specialized hardware for proving or verifying specifically um and so that that's that's kind of another dimension of potential improvement um that that i think holds a great deal of potential and just on that note real quickly i'd kind of like to remind everybody you know many of many of you know this but not everyone may so you know thinking about the internet and http the transition from mostly http websites to https um you know internet

encryption broadly speaking was only made possible when intel introduced the aes instruction set you know into basically into their hardware so that you know having having this stuff baked into hardware i think is a goal that um is is very exciting and potentially something that will really actually get to the point where you know an average person using a phone might be able to take advantage of this technology and use it to its full potential so um yeah okay so with that so what is the z price so okay so i just told you about how complicated and a little bit thorny of this problem that it is but so how do we come up what's the idea that we're pursuing to try and solve it um so first of all like i said accelerating zero knowledge cryptography means many things to many people

so the first thing that is very important i think here is to include as many voices as possible so z prize is an industry initiative i really want this to be something that we can all own and be excited about um it's not just it's not the end it's not the initiative of any one company or team it's it's all of us i think share this goal of making this technology practical and promoting it to the broader world which i think many folks don't really understand even the potential applications that this technology holds so so again the first the first thing to say is that it's all of us um the second is that the format to me that was most exciting um was this competition format that let me say a little bit more about so the x prize which some of you may be familiar with

uh early 2000's competition that is credited with making commercial space flight practical largely right so before in the all the way up until the late 90s the purview of human space flight was basically just in government space agencies and the x prize kind of famously made it a challenge for uh commercial space flight you know to to prove its viability and so i really like that model and i think that you know that's that's proven successful and would love to apply that here in the case of zero knowledge cryptography so the model is that we're considering is a competition where teams will submit their given solutions to the problems which i'll talk about in a minute and the the winning solution jumping all the way to the bottom um not only will the winning solution you know earn a prize

that potentially would be very interesting then but also the winning solution be open source right and i think that's that's the other really important piece of this is that you know this technology i think is too powerful and potentially transformational to leave inside of any one company or organization i think it's very very important that we share you know potentially the results of this competition and the results of these efforts collectively because as we've seen here with numerous proof systems numerous papers numerous protocols the the composability of different things is one of the most powerful aspects of open source software one person starts with one thing and then you kind of it one another team iterates on it and integrates on it and that's really how you get to um you know i think to the ultimate

goal um cool okay so the the goal of the competition at a high level is to you know accelerate and improve the baseline performance of key general algorithms and i say general because you know again i mentioned a minute ago i care very deeply that this is an industry-wide competition but everybody's working on different things right and uh you know so how do we basically balance the needs of individual teams versus the needs of you know versus the needs of kind of everybody or the interests of everyone and so the idea is to focus on some slightly lower level primitives that might apply in many cases right so many fry based proof systems for example rely on ntt or number theoretic transform operations so accelerating those potentially would

have an impact on not just one protocol but potentially everybody that used a fry based system similarly elliptic curve based systems you know marlin planck etc you know are very uh bound by multi-scale and multiplication operations as some people have pointed out earlier today so potentially accelerating that part of the process could have a great deal of impact and be more widely applicable and i kind of talked about some of these other things already but again i think the goal here is to build a foundation of libraries or implementations that other teams can build upon and iterate upon and then make you know potentially practical for wide for more widespread use um yeah so anyway so the z prize you know consists of two divisions the open

division is what i just talked about a second ago which is sort of really more of the things that are of general interest um you know and this has been something that now we have up 32 teams across the industry that includes many people that i see in the audience who have contributed expertise and and uh and feedback into selecting these given prizes which i'll show in a minute um but these are these are sort of the open division of the z prize is sort of intended to be for public goods so you know people can propose prizes and these prizes are things that are of general interest or usefulness and and then teams will hopefully compete to you know produce solutions for those that will be open source for the benefit of all um there's also the team division and the team division is really meant to capture

things that you know maybe are very specific to a given protocol or project right and i think the z prize you know again we have to balance here between the needs of the many versus the needs of the few right and you know the reality is a lot of expertise and potentially talent uh resides in a given project with maybe a given specific set of use cases that they care about or a given set of requirements and so the z prize should be flexible enough to basically support um you know a competition around just those maybe more limited um those limited prizes and that even for for teams that may not care about a more narrowly scoped prize i think the reality is as we've seen over the past few years um the space moves incredibly quickly so as

you know some other folks have pointed out earlier this morning like potentially as fry becomes more practical more and more teams will go to use that and so you know it's still good to encourage kind of specific optimizations for specific problems as well um so again so where are we um so i'm pretty happy to say that um so far we've made a lot of great progress and again many of you in this room are teams that are supporting this initiative um which now at this point we've raised seven million dollars in total prizes uh including from you know and we've got support from amd as you can see here it's gonna be providing fpgas for for one of the prize categories and uh and yeah and like i said so that's um 32 sponsors uh seven million dollars worth of prizes and we're excited to get uh to launch that here

are the open division prizes um that we have planned and the competition will be launching in june so we're we've now defined all the prizes raised money kind of gotten some outreach and awareness and then the next step is really to specify in more detail these prizes and these individual prize categories and recruit recruit competitors uh to compete um and so again open division here being more for general interest kind of use cases and so you know msm ntt acceleration across a variety of hardware platforms server which could mean either fpga or kind of commercial grade gpu uh a wasm you know accelerating these operations in wasm specifically which is of interest to many teams that care about you know having people perform these uh you know

do proofs or verify proofs on commodity hardware mobile um you know that is a category that i think not a lot of people have explored optimizing but holds a great deal of potential and you see little names of companies underneath each of these and so these companies that are that are kind of listed here have already volunteered time and expertise and effort into specifying the prizes specifically so like what is the you know how is it going to be benchmarked what is the platform on which it will be run and you know how what are the details that are relevant to the individual competing teams um so yeah so that's the this is the concept we have these are the prize sponsors but just because these names are listed here these are not the only sponsors of these prizes i'll go through the full list in a minute um but you

know but these are you know these are the teams that have volunteered to put forth uh the greatest effort for for kind of helping specify this uh and then we have the team division which again i think is more specific to given projects or you know ecosystems and so we have a large number of folks who are interested in this as well again a little bit more specific but potentially at some point in the future it could be more applicable to uh to all of us okay so timeline so some of you might remember i actually pre-announced this like a few months ago at zk hack or actually i guess was a month ago and this is you know today i'm talking about it and we we had you know some coverage and venture beat as you just saw and so this sort of constitutes the

official announcement and from here you know the next couple of weeks and several of you will be help over the next couple of weeks hopefully you know me and several of you will be working on specifying the specific prizes and the competition and then sourcing potential competitors and applicants um from hopefully not just in the zero knowledge space but also from outside so we can grow the industry of people or the industry and the group of people that is is able to submit solutions to these problems and benefit that would be to the benefit of all so that's starting in on may 1st um the applications will open up so if anybody knows anyone who might be kind of interested in these competitions and

i didn't show the actual prize amount let me ju let me go back here so i'll uh i can share afterwards with anyone who's interested or you can go to zprice.io to see the specific prize amounts but for example the server msm ntt acceleration for server has two million dollars worth of prizes associated with so you know basically all of these prizes have either 500 000 roughly you know anywhere from five hundred thousand to two million dollars worth of prize money associated with them so potentially could be quite enticing even for folks who are relatively specialized with high you know they have day jobs and maybe you know high opportunity cost for their time but um the idea is because we've raised you know seven million dollars we can we can afford to kind of entice those folks to uh yeah to come and compete so anyway that's over the next

month we'll be recruiting competitors and then june first the competition will start and we still a little bit undefined for how long we'll let it last i mean some of these k in some cases or well rather for certain hardware platforms things are a little bit easier to uh to do and and or harder to do so for example you know writing or optimizing approver for planck on a gpu versus an fpga you know the timelines might differ on on how long we need to give the competitors to uh to submit a good solution but roughly speaking the plan is to start on june 1st and on september 1st and we'd like to do some kind of award ceremony and hopefully announce um the winners of these uh of these different prizes and then have that again all available in open source um

potentially as part of an existing library or get and include an existing library potentially like artworks that everyone can then use um okay so i think i'm out of time but just quickly running through just want to say thank you to all the sponsors who are making this possible and to all the people who have contributed again feedback and and thoughts into how to make this uh make this competition run and lastly uh how can you help how can you get involved uh come find me after this talk let me know um follow zpr zprise.on twitter we're getting ready to kind of retweet the venture beat article and some other stuff so you can if you want to help amplify that that would be great um and then if you would like to either sponsor or compete please reach out let me or one of my

team know and we'd love to include you thank you very much [Music] you