00:00hi and welcome to the a 16z podcast
00:02we've talked a lot about crypto on the
00:04podcast you can check out the blockchain
00:06and Bitcoin tag on our website for more
00:08episodes but this conversation moderated
00:10by general partner Chris Dixon
00:12summarizes the bigger picture and
00:14overall trends in crypto currencies
00:16featuring past guests Juan Bennett
00:18founder and CEO at protocol labs and
00:21Olaf Carlson we CEO of poly chain
00:24capital it was recorded at our most
00:26recent a 16 Z summit event in November
00:282017 can you tell us how you first got
00:32into all this stuff so in June 2011 I
00:35found out about Bitcoin and just went
00:38down the veritable rabbit hole which is
00:40something people say in this space where
00:42once you start thinking about it nothing
00:44else feels interesting anymore so that
00:46happened to me pretty aggressively I
00:48wrote my undergraduate thesis on Bitcoin
00:49I joined coinbase as a first employee
00:51and worked there for three and a half
00:52years after that I started poly chain so
00:56poly chain is an investment fund that
00:58invests exclusively in cryptocurrency
01:00protocols we do not invest in the equity
01:03of private or public companies at all
01:04it's this area of kind of new protocols
01:08and icos and all those kind of words
01:10you've been hearing that's what we
01:12invest in one I studied the Soviet
01:16systems at Stanford and at the time
01:18there was this peer-to-peer winter
01:20innocence so later on and around 2013 12
01:24and 13 I started working on a
01:26distributed file system called ipfs
01:28which is now a large open source project
01:30and through that ended up discovering a
01:32whole bunch of stuff around Bitcoin and
01:33the thawing of the peer-to-peer winter
01:36and like the coming of the peer-to-peer
01:37summer again and that's when I got very
01:40heavily into all of this created the
01:41beginnings of falcone then then started
01:43protocol labs in 2014 and so protocol
01:45apps is a research development and
01:47deployment lab for networks like these
01:50we make ipfs file coin and some other
01:53protocols file coin is a decentralized
01:55storage network that uses a token to
01:57mediate the exchange between storage
01:59providers and clients and the point
02:01there is to rewire the proof-of-work
02:04function of a system like Bitcoin to
02:07cause useful storage so imagine all of
02:10the work that is being expended to
02:12maintain the Bitcoin ledger
02:13have that same kind of exponential
02:15growth rate to it to this competition
02:16but apply that to creating a useful
02:19cloud storage service that anybody can
02:21join so a lot of people in Silicon
02:23Valley think why do we need
02:24decentralized protocols like this you
02:27know we have these amazing super
02:29computers in our pockets our internet
02:30connected and great search engines and
02:32they're free and then sometimes people
02:33say oh it's sort of for government you
02:35know to get around the government or
02:37something like this and this is the only
02:37purpose of these things I would say that
02:39there's a couple things one is these
02:41things in volume are not actually free
02:42so if you try to store hundreds of
02:45terabytes or petabytes in Amazon s3
02:47you're gonna have a bad time just
02:49thinking about the bandwidth alone if
02:50you actually want to access all that
02:52data it's gonna be very very expensive
02:53so that's where there's a margin that
02:55you can cut out so there is something to
02:57be said around these systems can create
03:00a better optimization if you allow
03:02everybody on the planet to participate
03:04as providers in this network folk win
03:06similar sort of analogous to Airbnb
03:07versus hotels exactly you're taking
03:10advantage of excess capacity yeah I
03:12think that's the perfect analogy because
03:13there's a ton of spare storage on the
03:16planet that can't be added to our
03:19storage provider because it's extremely
03:20difficult to build a storage business so
03:22if you have a petabyte of storage great
03:24but how are you gonna sell it how are
03:26you gonna build a business how are you
03:27going to attract customers are you gonna
03:29compete with Amazon the bar is way too
03:31high for you to be able to sell those
03:32bytes on the network but if you had an
03:34algorithmic market that you could trade
03:36that storage and then the thing gets
03:38very very different right this is
03:39precisely like the right way of to look
03:41at it an Airbnb for storage where it's
03:44mediated by a token the other thing I
03:47would say is decentralization is an
03:49important security feature that right
03:51now we are taking we're not really
03:53realizing how important this kind of
03:55thing is so when you trust a huge third
03:58party and you trust them entirely with
04:00all of your data and so on they control
04:03everything around those bits and if they
04:05control those bits they effectively
04:06control your digital life and this may
04:08be fine to a lot of people who traded
04:10their digital life for you know really
04:12nice things like maps and the social and
04:15messaging and so on it is less so fine
04:18for businesses for governments and so on
04:19and so there is a very strong interest
04:22in uprooting from these free but not
04:25quite free clouds and to
04:27creating something that is verifiable so
04:29you shouldn't have to trust a third
04:31party to do the right thing you should
04:32be able to verify that they're actually
04:34doing the right thing and be able to
04:36move to someone else when they're not
04:37well we're seeing this with sort of the
04:39you know the regular hacks Equifax the
04:41DNC hack right like more and more people
04:43are realizing the clause in the
04:45centralized security architecture one
04:47key feature of these protocols is the
04:48idea that the users the token holders
04:50control the protocol are the owners so
04:52the file coin protocol right actually
04:54isn't really owned by one right the
04:57falcone protocol is a pure peer-to-peer
04:58system so if you want to invest in the
05:00file claim protocol the only way to
05:02invest in that is to buy file coins
05:03there's no company you can really back
05:05per say to get exposure directly to that
05:08protocol and broadly I think part of
05:10what one is referring to is a larger
05:11trend which is turning networks into
05:13peer-to-peer marketplaces so when you
05:15think about bandwidth CPU GPU cycles and
05:17storage these are the types of things
05:19that have historically been brokered by
05:22centralized entities but now you could
05:24have equally like a file coin you might
05:26someday see a Wi-Fi coin or you might
05:28see a compute coin where people can
05:30actually send out you know or share
05:32rather their cpu GPU cycles and earn
05:34money in a native token it's a fully
05:36peer-to-peer network and it just removes
05:37any inefficiency to the middleman now if
05:39you take that to its logical conclusion
05:40this basically makes sense for any
05:43extractor from a network many of the
05:46major web 2.0 companies that have seen a
05:48substantial growth in the last 15 years
05:49this is Twitter Tumblr Etsy eBay uber
05:52every single of these companies could
05:54someday see competition from a pure
05:56peer-to-peer protocol so instead of
05:57using Twitter the company you use
06:00Twitter the protocol and so there's no
06:02more value extraction by that
06:04centralized entity and instead that
06:06whole period of your network Twitter the
06:07protocol is actually owned by the users
06:10themselves now where do you think we are
06:13if that will somebody be the case like
06:14where do you think we are in that kind
06:15of evolution because it seems like a lot
06:17of the things being built today are
06:18things like what you're building Wan
06:19right which are more kind of
06:20infrastructure layer I've been sort of
06:22building kind of the equivalent of like
06:23the AWS for the web 2.0 era and not
06:25necessarily be a player so a lot of the
06:27protocols that are going to be able to
06:29do that kind of like a Twitter protocol
06:30type structure will require a different
06:33framing of infrastructure in order to
06:35operate in this nice peer-to-peer
06:36publicly verifiable context and to do
06:39that you need a whole bunch of
06:40at first so that's where we are today
06:43once those layers are structured then
06:45you can build a lot of the consumer apps
06:46there'll be a few consumer apps
06:47appearing in dozen 18 and so on but I
06:50would expect to find actual competitors
06:53to the world of normal social where
06:55people are choosing these applications
06:56for their own right not because they're
06:58in this different world later like maybe
07:01late 2018 or 2019 so I think this this
07:04period is entirely about or mostly
07:06focused on building out the
07:07infrastructure to power those
07:09applications later on yeah and one way I
07:11think about this is what I call the web
07:143 technology stack so this is instead of
07:17the web stack that you'd see a website
07:18on and so this is built on all the
07:20technologies we referenced earlier like
07:22AWS and things like that this is the
07:24stack of technologies that you need to
07:26run peer-to-peer applications so going
07:27back again to the peer-to-peer Twitter
07:29there's this really simple problem today
07:32which is where do I store the tweets
07:34right it's a super baseline problem but
07:37until you have ipfs and an incentive
07:40system using something like file coin on
07:42top of that you can't really have
07:43reliable pure to pure file storage so
07:46you can't really build twitter of the
07:47protocol until file coin is deployed and
07:49robust and easy to use and we're
07:51definitely in the infrastructure
07:52building we are in the phase we're like
07:54with the normal internet you had to kind
07:56of log your IP address and things like
07:57that I don't think we really even have
07:59the browser maybe so I think it's early
08:02but the consumer apps will come in you
08:04know call it three years or something
08:06like that so it's a point what are some
08:08of the projects that you're tracking
08:09closely anybody know a lot of these are
08:12mostly developer facing they're not
08:14really meant to be used by regular
08:17people they're really not so for example
08:19ipfs and file coin could be used
08:21extensively by regular people in the
08:23same way that when you tweet that gets
08:25stored on Twitter servers but you don't
08:27really think about mechanically how that
08:29happened so there's a layers of
08:31abstractions that will be built on top
08:33of most of the things that we're
08:34investing in today so most of what we're
08:36doing are like they're you know to use
08:38like a lot of lingo would be a
08:39verifiable off chain computations so
08:42this allows smart contracts in smart
08:44contract systems like aetherium or Tazo
08:46sword affinity to actually have more
08:49complicated logic that's computed off
08:51the blockchain and then a proof
08:52is sent back to the blockchain in order
08:54to show and prove to the rest of the
08:57network that that computation was valid
08:58that's a system called true bit but this
09:01is all as you can probably tell not true
09:03bit is absolutely not meant to be
09:04user-facing it's a developer tool so
09:06that developers can build things that
09:08become easier facing over this is just
09:10to maybe get some context so you know we
09:12think as we described there's a lot of
09:14benefits decentralized systems one of
09:15the big disadvantages is that least
09:16currently is scaling so aetherium can
09:19what handle like ten transactions a
09:20second or something which is just many
09:22orders of magnitude lower than a system
09:25like Facebook right and so there's a
09:26variety of efforts on many different
09:28fronts to make those system scale better
09:30and like true--but is one way to do that
09:32and so there's a lot of stuff going on
09:33now that we're all involved with which
09:35is how do we get these systems to scale
09:36better which i think is very similar to
09:37what happened 195 to 2010 sort of on the
09:41infrastructure side of the kind of web
09:422.0 world right yeah I think it's
09:44exactly right the cloud went through a
09:45whole bunch of phases of optimization
09:47and that's where we're hitting so
09:48Bitcoin introduced block chains and this
09:52open ended consensus system where you
09:55could add any transactions but it was
09:56horrendously slow and just like the
09:59throughput is abysmal for what you
10:01actually need in an application right
10:03and so it's partly for that reason I
10:05think that Silicon Valley effectively
10:07dismissed a lot of the potential of
10:09Bitcoin it just looks like such the
10:11wrong thing if you could you know beat
10:13the through protip Bitcoin with a secret
10:14light database in your phone why is this
10:16thing useful and so everybody kind of
10:18missed the point the point was to
10:20establish this different kind of
10:21computing context where you have public
10:23verifiability and immutability and once
10:26you show that that's possible then you
10:27apply all the standard techniques for
10:28scaling this kind of thing parallelism
10:30hierarchy and so on and you can build up
10:33something that's really high throughput
10:34by either moving the computation off
10:36chain by having faster block chains and
10:38all kind of stuff so expect that to
10:40happen over the next year or two
10:41already there's a bunch of academic
10:43results that show this is not only
10:44possible but it's provably so so now
10:47people are building those
10:48implementations and will see them live
10:50in the next two years or so so let's
10:52look what I see is people who probably
10:53heard about ICS there's good and bad
10:55obviously mostly bad so this concept of
11:00an ICO though theoretically is very
11:02interesting so this stands for initial
11:04coin offering and this is the
11:05concept that when I want to launch a new
11:07cryptocurrency protocol or new token
11:09built on top of one of these smart
11:11contract systems like etherium I
11:13actually raise capital from kind of the
11:17crowd on the internet so anyone in the
11:18entire world in cryptocurrency so like I
11:21literally raised Bitcoin and raised
11:22etherium in order to distribute my new
11:25cryptocurrency token so part of the
11:28reason this is so theoretically
11:29interesting for me is that for the first
11:31time we're seeing cryptocurrency
11:33technology being used to accelerate the
11:35development of cryptocurrency technology
11:37so I think that despite ICO is right now
11:40being very noisy sort of scammy and all
11:42these things it has created an
11:44inflection point where when you have a
11:47brilliant idea the capital is much more
11:49readily available from a large group of
11:51people that actually conceptually
11:52understand what you're working on and so
11:54the kind of literally billions now that
11:57have been raised in icos just in the
11:58last year it has created an inflection
12:01point for this entire space where the
12:03pace of development will never be it's
12:06never been faster and I don't think it
12:07will ever go the other way because the
12:09Gini is sort of out of the bottle
12:10it's crowdfunding done right in a sense
12:13in that you now have an infrastructure
12:16project that you can actually crowd fund
12:18from its users and developers and
12:20investors who understand the potential
12:22value of this thing and they're funding
12:23it and you're going broadcast straight
12:26to the whole world and making a case for
12:27it there's so much capital right now in
12:29the world that that's why you're getting
12:31this massive level of noise right so a
12:33ton of people are flooding into the
12:35space that don't really have a good idea
12:38of how to tell the difference between a
12:39really solid project and one that's not
12:41and that's what creates a whole bunch of
12:43noise the moment that a lot of projects
12:44were able to race a ton of capital very
12:46easily then that attracts a whole bunch
12:48of players that are not that good and
12:50can kind of pass off that they're good
12:52and so that that's what we're seeing
12:54right now we're seeing probably like
12:55hopefully this is the peak of the hype
12:57cycle hopefully this will not get worse
12:59next year so hopefully we have now
13:01peaked this and and it's going to go
13:02back down and then suddenly we'll get a
13:04lot more rigor in investing so I think
13:07things that would help the spaces to
13:08establish very credible groups that can
13:11analyze the stuff and provide
13:13independent reports let's say this is
13:15why this thing we think this thing works
13:17or this is why we think this thing is
13:19so that people can understand that
13:20discourage people from investing in
13:22things that that they don't fully deeply
13:24understand why it's gonna work I don't
13:26know how well we can do that but well
13:28that's one thing and just setting
13:30standards as a whole community of saying
13:31projects should have actual degrees of
13:34implementation really strong teams
13:36certain levels of track record before
13:38going into this thing right so that'll
13:40take curation right so it'll take a set
13:41of investors and and platforms and so on
13:44that can apply a certain level of
13:45curation to the entire system to distill
13:48the signal from the noise people hear
13:49about the news like Bitcoin and you know
13:51the debates going on there
13:53it I think it tends to be something
13:54that's focused on much more by the news
13:55and not by people that actually work in
13:56this space yeah I think it brings up a
13:59larger point which is governance of
14:01these protocols so this is the idea of
14:04you know how do you decide what to do
14:06with bitcoins the protocol so once this
14:09thing is live and running it's like a
14:10live software project live financial
14:12system and then somebody comes up with a
14:14way to improve it that may have
14:16trade-offs may not have trade-offs just
14:17like how do you change the protocol and
14:19right now in pretty much every live
14:22system there isn't a clear governance
14:25there's no formalization around how
14:27decisions are made or executed it's
14:29really all very vague and I think one
14:31thing that's very interesting to me is
14:33on chains like on blockchain governance
14:35systems where you actually use the
14:37protocol to create a voting mechanism to
14:40actually upgrade and change the protocol
14:41there are a lot of projects working on
14:43this right now but if you can do this
14:45effectively it basically means that the
14:47protocol governs itself right and now
14:49there's no entity outside of that
14:51protocol that has undue influence on
14:53that protocol and also presents a
14:55weakness both from a security and
14:56compliance perspective so to me for
14:58these protocols to become more robust
15:00and also iterate a lot faster on the
15:01technology we're going to need on
15:03blockchain governance systems where
15:05people can actually stake money in order
15:07to vote on protocol changes which then
15:10everyone can kind of agree on to move
15:12the protocol forward completely agree
15:14that where governance is right now kind
15:16of sucks hey Corinne I think is the best
15:17example of this the debates that have
15:19been raging for the last two years are
15:21about relatively minor things so
15:24increasing the block size became this
15:26huge holy war and when you step back and
15:28look at what they're talking about is
15:30just increasing one factor
15:32right you know 2x or something like that
15:33and which academically people proved
15:36we'll have close to zero impact on the
15:38actual network so this is a huge holy
15:40war around people not agreeing on some
15:42very basic facts and so absolutely we
15:45need really strong governance structures
15:46I think a theorem has done a lot better
15:47in that regard but it's still very
15:49unclear I also think systems like either
15:52on blockchain governance or some kind of
15:55guidance mechanism will work then again
15:57we should also look back at the most
15:59successful open-source projects and how
16:00they do it Linux is run by Linus
16:03Torvalds this is like one person that
16:05ends up delegating to a whole bunch of
16:07other people to prove why it's
16:08successful not arguing and that's
16:10necessarily the best thing but as we go
16:12into building the future and building
16:13new things we should also figure out
16:14what has worked in the past and what
16:16hasn't people did try to come up with
16:18voting systems for a lot of open source
16:19and it ended up going worse so there's
16:21some open source projects that were
16:23running to the ground effectively by
16:24just communities of people effectively
16:27using democracy to kind of drive the
16:28profit into building way too much trying
16:31to be way too much for a lot of people
16:32so there's a an element of curation
16:33there that it's important to not lose
16:35there's another factor that is gonna
16:37become relevant with proof of state
16:38protocols and if we end up building this
16:40on chain governance thing which is you
16:42don't necessarily want majority of money
16:44to decide the outcome of the protocol
16:46and every decision because that could
16:48lead to really bad incentives so if you
16:50just allow you know pure majority of
16:52money vote in say the US governance
16:54which is not I guess not that different
16:56from what's happening but there's a very
16:58strong difference there where at the
17:00very least you have a whole bunch of
17:01rights and laws in place to prevent just
17:03arbitrary decision making the majority
17:06of ether for example can't vote to say
17:08suddenly destroy all the other groups
17:09right or something like that so I think
17:11this is very greenfield stuff but it's
17:14gonna impact the scalability and utility
17:17of these protocols down the road and so
17:19it's a big problem to solve so if people
17:21have good ideas around this you should
17:23build something just explain how much
17:25electricity bitcoin mining uses this is
17:28exactly like what one is trying to do
17:29here is is take that and make it into
17:31something that's actually useful like
17:32storing files instead of just wasting
17:33electricity and it's also what a lot of
17:35people are working on like aetherium is
17:37switching over something called proof of
17:38stake which is going to hopefully solve
17:39that electricity wasting problem so when
17:43the news articles that say that all the
17:44cryptocurrencies waste money like this
17:46is actually one of the main efforts in
17:48the community is to make more efficient
17:50use of resources yeah it's kind of
17:51insane how much power bitcoin consumes
17:53it's about 1/2 well as incentives yeah I
17:56mean I'd like it over it over shot in
17:58ways yeah last we calculated it was more
18:00power expenditure than all of Slovenia
18:02and so I think it's much more now
18:05because of the recent spike a lot of
18:06miners came in in the last couple of
18:08months so it's a lot of power and so if
18:09you turn that into useful work then you
18:12get all of this additional side effect
18:13of it the good news is are the good or
18:15bad news it shows you the power of the
18:17incentives of these systems Satoshi is
18:18white paper which is basically just a
18:19set of incentives and they work so well
18:21arguably too well in the case of mining
18:23that it literally created this much
18:24electricity waste but it shows you the
18:27power of these systems and what if you
18:28could harness those kinds of incentive
18:29systems for something better yeah we see
18:31that's one of the most important things
18:32for the entire space so the proper
18:34application of game theory into these
18:36systems so I think computing has had
18:39game theory in it for a very long time
18:41but never quite like this never to the
18:43point where you can just describe a few
18:44incentives and code them into a protocol
18:46and then release it to the world and
18:48just watch the effects happen
18:49watch massive mines appear out of
18:52nowhere and start mining Bitcoin so that
18:54is extremely exciting because there's a
18:56lot you can do with incentive structures
18:57to turn normally centralized systems
18:59into decentralized ones and build a
19:01whole bunch of applications that will be
19:02very difficult for one small party or
19:04one single party to achieve on its own
19:06yeah and one thing about these systems
19:08broadly is that ultimately a blockchain
19:11protocol is just a way to organize
19:13humans very very precisely and without a
19:16central hierarchy telling people what to
19:18do so you could think about like if
19:20Bitcoin was a corporation trying to
19:21organize the mining it would never be as
19:24efficient as it is with just this
19:25decentralized protocol that incentivizes
19:27all these various actors around the
19:29world to do this very precise behavior
19:30and you see also very very efficient
19:33capital coordination in a lot of these
19:35systems so like with icos
19:37I think that it's probably the fastest
19:39and most efficient capital coordination
19:41we've ever seen in capitalism like since
19:43money was invented people literally put
19:45a cryptocurrency address on a web page
19:47and within 30 seconds raised a hundred
19:49million dollars so what are some of the
19:51things you guys are excited about the
19:51next couple years I'm really excited
19:53about the application of smart contracts
19:56are more advanced logic that can be
19:58built into a blockchain like aetherium I
20:00think that lots of exciting things while
20:03we built with smart contracts over the
20:05next couple years that are sort of hard
20:07to imagine more like sort of Sakai just
20:10because it's such a new tool that I
20:12don't think the world has ever really
20:13seen I would say in the short term
20:15things like formal verification coming
20:17to smart contracts so we don't have
20:19single line bugs that lock up hundreds
20:22of millions of dollars whoops in the
20:25kind of medium term once the
20:27infrastructure layer is structured
20:28then what are those applications that
20:31are going to come like what are the
20:32consumer apps that are gonna be so good
20:34on this medium that have nothing
20:36necessarily to do with decentralisation
20:38or crypt assets but are inherently a
20:40really good consumer application because
20:42that's what's gonna make all of this go
20:44mainstream and I think in a long term is
20:46just rewiring power structures on the
20:48Internet I think there's the possibility
20:49here of enabling keeping a very open
20:53Internet in the long term which is not
20:55necessarily going to happen so the
20:57freedoms of the Internet have been
20:58eroded for a while and they're going to
21:01continue being eroded especially there's
21:03any kind of bigger geopolitical problems
21:05than potentially the internet might
21:06charge like who knows right there's a
21:08whole bunch of censorship and
21:09restriction already and it could be that
21:12we don't have end-to-end encryption from
21:14one country to another in the future
21:15like there could be a sequence of events
21:16that happens it prevents that and so I
21:18think that these systems can enable can
21:21build their foundations on an incentives
21:22layer to make sure that we preserve
21:24those kinds of freedoms and so I think
21:25that's that's a very I think this is
21:26already being its approach which one of
21:28your protocols is already being used for
21:29anti censorship yeah so catalunya try to
21:32declare independence and try to vote for
21:33it the Spanish government censored the
21:36website and so then the Catalan
21:38government put their website on ipfs and
21:41and distributed that way so it's pretty
21:42cool to see the Catalan president
21:44tweeting out the link to ipfs so that
21:46was exciting so we're seeing the first
21:48of all these of this kind of thing
21:48that's gonna increase over time but
21:51it'll come to a head at some point when
21:53governments sort of governor said you
21:55know in a big way catching up to the
21:56freedoms of the internet and figuring
21:57out how to restrict them and how to deal
21:59with them in some places enable and in
22:01some other places restrict them so think
22:03there's gonna be a ton of questions that
22:04are gonna hit hopefully not next year
22:06probably later but hopefully by that
22:10really strong infrastructure to remain
22:11free all right great thanks thanks Chris