Working Towards Permanent Information Storage with Sam Williams, CEO & Co-founder of Arweave -Ep.30

Welcome to the CoinGecko podcast. For 
today's episode, we have Sam Williams,   CEO and co-founder of Arweave, a decentralized 
storage startup. So very happy to have you on the   show. Welcome to the CoinGecko podcast, Sam.
Thanks for having me. It's good to be here.  Yeah, to start things off, Sam, maybe can 
you give us at CoinGecko,the community here,   a simple explanation of what Arweave is 
and would like to know more about it.  Yeah. So Arweave is a permanent information 
storage system, I guess your audience is very   aware of blockchains. And so they're probably 
aware of the idea of content storage. Storing   something in the blockchain itself.

We 
can think about Arweave basically as   on-chain storage that scales and has appropriate 
incentives to me that were literally permanent.   So the way it does that on technical side is 
a new mining algorithm, which rewards people   for storing data, rewards them more for storing 
data that is stored less. And that kind of auto   levels the storage over the disc space that is 
provided. And then on the economic side we have,   what is essentially an endowment structure. 
So you paid for essentially 200 years worth of   storage upfront when you use Arweave, which sounds 
expensive, but it's about 0.40 cents per megabyte,   approximately. And then over time as the 
cost of storage declined to essentially gain   interest in the form of storage purchasing power 
on that initial 200 years. Yeah, the grace period   you're using it essentially. Just like with a, you 
know, university endowment, you make some interest   on the principal that's put aside and you use 
the interest to pay for people to go through,   you know, get degrees. Yeah, it's the same 
principle, but just applied to data storage.  So, yeah, I kind of get the idea, where instead of 
paying for one year or one month, you pay for 200   years upfront and then you got to store your 
storage essentially, almost permanently.   I guess my question is, how do you come up with 
the 200 years number? Like why 200? Why not 100   or 500 and I guess this is too far to ask but what 
happens after 200 years? Like do they disappear?  Right.

Exactly. So over time, the cost of 
storage declines, and you only take from   the storage endowment, what you actually need 
to pay for storage. So essentially what you're   getting is interest on that 200 years in the form 
of storage purchasing power over time. So imagine,   and this is the average rate, the cost of storage 
declines at a rate of 30% per year. That's what   it's been on average for the last 50 years.

So, 
year zero, when you put it in, we've got 200 years   with the storage purchasing power up front. At the 
end of the first year, now we actually have 260   minus one years worth of storage purchasing power 
for the principle that we've already put up. And   you can see as this compounds, you actually end 
up with more storage purchasing power, not less.   So the question of why 200 and it's actually a 
reciprocal, which is why 0.5%. So why expecting   a rate above 0.5% for storage cost to clients? 
It constantly comes down to conservatism. Like   we think that blockchain protocols are the 
successful offer a zero to one service,   something like Bitcoin, for example to digital 
scarcity from zero to one. Ethereum took smart   contracts from zero to one. These are not just 
copying business models that you can do in a   centralized world in the decentralized world. 
These are giving you the ability to do something   you just couldn't do before. And we think that 
permanent storage is another one of those.  It's something that you couldn't trust a 
centralized company to give you at any point   in time.

You can however, trust a protocol to give 
you this service because it won't change. And the   participants can come in and out over a very long 
period. And there's no corporate governance to   worry about. There's no profit motive or incentive 
for the network exactly in and of itself and   so on. So it's the kind of thing that you can 
allow a network to do, but in order for it to   takeoff, you need it to be sufficiently 
conservative.

And when you say, geared towards   achieving it's one fundamental goal for people to 
be able to trust it. So we could have probably set   it at 20 years and that would probably be okay. 
But if we'd done so then people might not trust,   it might not catch on. You might not actually 
about to achieving one core important, you know,   zero to one, if you will, change that it allows. 
So we picked a much more conservative number.  I guess the question then is if it's 
conservative, it almost feels that if I want to   store something on Arweave now, I'm sort of 
overpaying for the endowment fund.

Right.   And this cost of 200 years, someone's going to 
make a profit from this, right? So who is going to   take the profit from me overpaying on 
this endowment fund? Is it the protocol,   the miners or who's gonna take this profit here?
Yeah, it's an interesting question. Well,   at the moment, what happens, what will happen 
and I don't think this will ever change   is just the funds just go into the endowment, 
where they just stay. Forever, basically.  Okay. So the protocol has this endowment, 
which controls all this funds, I suppose.  Exactly. Yeah. So it's kind of like burning 
with the caveat, you know. [inaudible] if   for some crazy reason in the future, the 
storage costs actually increased then the   amount taken from you [inaudible] minus could 
also increase as well. So "burns*" if you will.   And [inaudible] I probably 
gets benefits token holders.  And the endowment, I mean, when I pay for 
storage on Arweave I pay them in a form of   AR token, right? And these AR token fluctuates 
with the USD price.

So if the price of Arweave   goes up in the future, that means, yeah, your 
endowment will have a lot of money and then…  That's already happened a lot.
Yeah. I've been monitoring [inaudible].  The endowment has like, I think at this point, 
maybe $250,000 worth, $300,000 worth of storage,   worth of USD inside it to pay for storage. And 
at the current time we're talking about like   five and a half terabytes or something, which 
in temporary storage terms not a lot, but in   permanent storage terms actually it's quite a lot. 
But even still, you know, now the expected storage   [cost decline rate], I calculated this recently. 
I wish I could remember the exact number, but it's   the expected storage cost decline rate to keep 
this to work perpetually at its current rate,   is something new or it's a 0.01% storage 
cost decline per year and then the thing   will run forever.

But yeah, there's definitely 
another reason that we chose 0.5%, which also   gives that reciprocal 200 years, is because 
it just gives a lot of slack for potential   changes in the token price, which we all know 
like early crypto networks, there are changes.  Do you see of any problem, like, when token prices 
rising, like everything works out well, right. But   touch wood, but like, what happens 
if the token price goes the other   way around instead then like [inaudible] 99%, 
for example? I hope it doesn't, but I mean,   if it does then will it still work out then?
Yeah. So I think 99% is a little bit extreme,   but like 90%, who knows, yeah.

That's why we use 
such conservative value. So that's just another   reason why basically. Yeah, this point, obviously, 
you know, there's so much in the endowment and   it's interesting, cause it goes both ways. 
So you put a lot of tokens in the endowment   and then the price moves up and now those 
tokens are worth even more, but obviously if   they go down and then you still have this extra 
buffer that you've generated. So now it's 0.01%   can afford to go to 0.5% and again or whatever 
it happens to be. It's an interesting question.  Cool. So decentralized storage is not 
essentially a new concept. I mean,   I've been in this space for some time and it's 
been attempted by several teams a few years ago,   since 2014 or so, or 2015, like Siacoin, 
Maidsafecoin, Storj and more recently   Filecoin with IPFS.

I think you kind of 
mentioned briefly that the differences between   this different decentralized storage is that 
Arweave is kind of going up to the permanent   storage solution, whereas I believe the others 
are not going after the permanent storage,   they go more for a fixed time storage 
solution, but I would love to hear from you,   what are the differences between Arweave and 
all the other decentralized storage solutions?  I think you put, what did they say, knock 
the nail on the head or whatever it's. Yeah,   it is a permanent storage system, which none 
of the others offer. And no one is actually in   this lane at all. Frankly I think it's kind 
of surprising because it's just such a big,   obvious and big, opportunity. I personally 
find that, so I was invested in the storage ICO   actually, way back in the day.

And I think like 
temporary storage system interesting. They're not   zero to one. Right. They're going to have to go 
and try and compete in Amazon on cost. And that's   going to be really, really tough, like really 
tough. I'm not sure that that's a winning strategy   for decentralized networks. I think you have to 
do something new and that's kind of why. Yeah.   Arweave is focused on this permanent storage 
angle. And then, also on top of Arweave, which   we haven't really covered yet, is what we come 
up with.

So it's like a collection of protocols   basically expose all the information in the 
Arweave to people's web browsers. And that's   like a whole different game. Like it's literally 
permanent decentralized web that lives inside   the system. Yeah. I don't think [inaudible].
[00:09:07] So how'd you get this idea to build   a permanent storage solution? Everyone's 
kind of building on this temporary or   fixed time storage solution, but you came 
about and say like, "Let's build something   truly game-changing, a permanent storage". 
Like no one's ever thought about it before.   How do you get idea? Why do you do it and all?
Yeah. Well if you've ever read 1984,   yeah, you know, the idea of the memory hole. Okay, 
for the listeners that haven't read the book,   basically the book is about a dystopian future or 
would have been the future, it's actually written   in 1948 and it was talking about the future as 
1994. And dystopian future in which authoritarian   regimes essentially around the world.

And they did 
so largely through the control of information. And   part of that control of information was controlled 
of access to information about the past.   And it's through this mechanism, they're 
essentially allowed to control the way people   think about the present and subsequently the 
way that they act in the future. It's just like   a really key insight. I think from that period 
of time is that actually how you control large   numbers of people is not necessarily through 
force. You do so through control of information.   And you know, this book was trying to exemplify 
that idea. And so the main character in the book,   Winston and also the smallest sub-unit of Arweave 
token is called the Winston, after this character.  The main character in the book, his job is 
to censor records of history by throwing   them down what he calls, the memory hole, which is 
essentially a fire pit. So his job is to collect   up all the records of a past historical event 
that is now unfortunate or the regime doesn't want   people to know about and to burn them all.

Yeah, 
we essentially wanted to make that impossible. And   we saw that the blockchain technology, you quite 
a lot of the way there, you got replication of a   piece of information all across the world without 
a central leader. And you've got the means through   which to add data to that. But of course, when 
you didn't have was the means to add arbitrary   amounts of data and that's the incentives to 
make that storage same and consistent. Like   people think at the moment, even if they do store 
data on Ethereum, some NFT people try and do this.   Yeah, like I know it's just crazy. It costs 
like thousands and thousands of dollars for 32   kilobytes, but whatever people try.

Yeah, there's 
no actual incentive for people to store that data.   The miners are only incentivized to store just 
the minimum amount that is required to take part   in the mining process. Even if it did blow up 
the size of the chain to store more stuff in it,   then people aren't incentivized.
So anyway, we figured, okay.   We saw the world moving in this direction where 
it would make more sense to have a decentralized   reliable archive of history. And we saw the 
kind of technological components that would   be needed to do that. And the first thing 
we set out to try and do is actually build a   newspaper archive.

Yeah, it was just basically 
trying to get newspaper articles and make them   live forever. So we didn't forget history, 
but of course we realized pretty quickly,   as soon as you solve those problems for 
a newspaper archive. Well, now you have   a new kind of archive you can use for everything. 
Yeah. and that's kind of the Genesis of the idea.  Interesting idea how you taught about this 
dystopian future and how governments will   try to censor things, and you try to build a web 
where someone in the future can sort of reference   things from the past and see for themselves that 
this is what it is.

Kind of like the internet   archive in a decentralized manner. Yeah.
Yeah. [inaudible]. Yeah. Interesting.  So you mentioned briefly just now about how the 
other decentralized storage solution were kind   of like competing against trying to compete 
against Amazon S3 and that's kind of like,   not a good idea because there's no way one can 
drive efficiency gains and compete against Amazon   S3. So maybe to give us some perspective of cost, 
I'm not so sure if you're aware of how much the   others are charging like how much it costs to 
store an IPFS versus how much it costs to store   on Amazon S3, and then how much does it cost 
to store permanently on Arweave, for example.  Yes. The others will be somewhat cheaper. I think 
people have done like the crossover calculations.   Like at what point does it become cheaper to 
store on Arweave that kind of thing.

It's like a   few years. But I really think that the benefits 
offered by permanent storage for information.   So being able to, for example, commit a blogpost 
and know that that blogpost, your ideas,   your writing can be perpetuated literally far, 
far past your death, like far into the future.   Presumably would forgotten after the last 
human or human-like intelligence has gone.   That's a pretty cool thing to be able to do for 
like what, you know, 0.40 cents per megabyte.   And for a normal blogpost. And this is 
an interesting question, like, what data   doesn't make sense for normal blogposts? That's a 
power that Arweave gives you for, like, you know,   let's see, if that blogpost 
is 30 kilobytes or something,   in 0.001 cent, someone's gonna check my Maths.
Yes, it's cheap.  With tiny, tiny fraction of a cent. Yeah, 
so that's actually a pretty damn good deal.   But of course, if you've got like, I dunno, an 
archive of really, really, really unimportant   video, something that's like a very, very large 
file format and you want to store a lot of it.

And   the density, what we call the information value 
density is low, basically. Then you might not want   to store it on Arweave. You might want to store 
it somewhere else. And that's totally okay by us.   Please go store it on Storj. My Storj tokens, 
which I still have or rather haven't picked up   yet. Anyway. Yeah, it makes total sense to use 
another kind of system for that. But there are   many, many things we would argue, almost 
everything that people publicly commit to the web   that they really want this component of like 
spreading it as far and wide as they can. If   only the functionality were available 
to them. And that's what Arweave does.  I'm just curious, right, for Arweave, like, 
do you have any stats on how much data that   is currently store on the Arweave network, now?
Yeah, it's under like five and a half terabytes   or something, but the difference is that 
this is permanent data.

They're like   replications of this far, far more than you would 
expect in a normal system. So I would say at least   until the like 250 to maybe a 1000 range at this 
point, quite possibly a lot more than that. Yeah.   But also again, it's about the information 
density. So I would say this one is probably   15 million, maybe 20 million different 
pieces of data on the network as well.  You mentioned information density, but I 
don't think I really understand what you   mean by information density. So maybe you 
can elaborate a little bit more about that.  So you've got a file that is one terabyte and 
it's value to you might be $5, right? You've got   another file, like this blogpost is 30 kilobytes 
and it's value to you is a few hundred dollars.   So you can take these, these factors and you can 
calculate it approximately a value density for   the data, if you will. Does that make sense?
So, when you store data, you can sort of put   a value on how much you value the data or…
It's an innecessary heuristic that people go   through in their head.

So it might not make 
sense if the value data density is low,   for you. You'll instinctively be like, 
I don't want to pay that price because   the files aren't worth that much to me, 
basically. But if it's something like a blogpost,   then the value density is much higher and you 
see like, yeah, that's like, you know, hundredth   of a cent or whatever it was, is a great price to 
store this thing forever. Do you know what I mean?  Okay. Yep. Yep. And I'm just curious, right? 
You mentioned about five and a half terabytes   of data that's been stored on Arweave right now. 
I'm just curious, so who are the guys that are   storing data on Arweave now? What kind of use 
cases have been shown to be using Arweave now?  Yeah. So one of the first ones that started 
getting picked it up most recently, well, I'd say   there's three key areas right now.

People starting 
NFTs. It's one that's growing excessively rapidly.   Like there's a real growing understanding that 
like, "Hey, if you've got an Ethereum contract,   which has got a URI in it, right, and that 
URI is stored somewhere that is not permanent,   then eventually inevitably, all of them will 
break. Like they are all stored on someone's   centralized server. And one day someone is going 
to turn that off and then eventually your NFT is   going to be worthless.

That's crazy. It doesn't 
need to be this way. Just do it on Arweave   and then it's there forever. It makes sense. 
People are starting to do this on mass very,   very rapidly, which is really cool to see. 
Another one is actually chain storage. So one of   the interesting prompts you have with blockchains 
is that they're essentially ever-growing ledgers   that don't have any incentive to store as I was 
describing before, actually. This is kind of a   problem. Eventually you get to this point where 
the chain is just too big. I was actually just   talking to someone, they're like, the Terrachain 
or something like this, I believe it's like 200   gigabytes and the software after syncing it is 
not necessarily so optimized for giving me that   number of transactions.

And so it's taking like a 
year to sync to the latest block. That's a bit of   a problem. So what you can do is you can store all 
that data on Arweave and you can have them nodes   very, very rapidly collate off the Arweave network 
where it's permanently stored, incentivize forever   and sync extremely quickly with the network.
So that's pretty useful. That started with SKALE,   I think were the first to do that. And then 
very rapidly after that, it was Savannah.   And then after Savannah, it was Polkadot and now 
there's like three or four more people we haven't   yet announced that have come on board to do this. 
And this has grown very rapidly into uh, projects   on top of the network, actually a startup called 
KAI.

And KAI, basically the idea is to build   permanent verified data feeds. So essentially 
they were just trying to solve this problem   with like we put together a co-funded grant with 
Polkadots where we're saying, "Hey, let's get all   the Polkadot data onto your Arweave". Make sense. 
It's not so complicated. And they're like, "Well,   what if the person that's running this bridge is 
actually malicious? How are we going to know that   they're not tampering with the data". Then they 
build have this really elegant solution using   SmartWeave, the smart contracting platform on top 
of our, to reward people for essentially running   these archiving nodes and verifying each 
other's data. And so now you can get these like   verified data feeds of almost anything. So that's 
really, really powerful. Pretty exciting to see.   The third component is people building what 
we're calling profit sharing communities.   This is something in the Permaweb ecosystem. 
So in the stack of protocols on top of Arweave   that expose all this stuff to people's web 
browsers.

I think this is super, super powerful,   and we're only seeing the beginning of it, but 
it's starting to take off very, very rapidly.  They've only been, this ecosystem has only 
existed for like six months. And I used to   count how many dollars have been committed 
to funding these rounds of that building.   It started as like, you know, half a million 
and it was a million very rapid, again,   it was 3 million, now it's 5 million and it's 
like, it's increasing at an exponential pace.   So that's pretty cool. But the basic principle is 
building decentralized web services. So services   top to bottom. They're running in a decentralized 
way.

You can imagine like a decentralized Twitter,   for example, where the community owns 
the profit created by the application   and the entire thing runs on decentralized 
infrastructure. I think one of the things   that is kind of missed in the crypto community 
sometimes is actually a lot of these dApps   that we use. They like decentralized one small 
component and everything else is centralized.   These are the opposite of that system. 
This is like taking it to the nth degree,   what if absolutely everything ran on like 
censorship resistant distributed infrastructure.  Interesting, the three main use cases 
that you mentioned. NFTs is like,   a lot of people don't realize this, like you 
mentioned like this the images are store on some   Amazon S3. And at some point somebody tried to rug 
some of these NFTs and they replace all the NFT   with actual rug images. That was quite hilarious. 
I mean, we minted some NFTs before this and we   were kind of studying the architectures, like what 
we found was that yeah.

I mean back then it was at   least kind of a pain trying to put onto IPFS, so 
we kind of, okay, let's just put it on S3, just   put it out and try out and experiment with NFT. 
But we see how this was unknown to many people   like, where is the actual image stored and I'm 
glad that people are talking about it right now,   but we saw this a couple of years ago.
Yeah. It's rapidly becoming, I mean,   we've been chatting about this for a long 
time for it's rapidly becoming like the   known thing in the industry.

There's this effect 
called like an information availability cascade,   where like it goes from suddenly almost 
no one knows about an idea, a few people   know it and then very suddenly everyone 
knows. It kind of going through that mode.   Yeah. Pretty interesting to watch.
Yeah. Interesting use cases on the nodes as well.   Cause like yeah, it is kind of, it's very painful 
to try to sync a full node and I can imagine that   moving forward, like instead of like syncing on 
your local machine, you probably sync it to a   decentralized Arweave note, which constantly 
gets added with new blocks and then you   can kind of [inaudible]. And then in the third 
one was the profit sharing community. I have   some questions on that, but keepI keep it for 
a bit more, but let me ask you this question   next. Do people store like private information? 
Like I know you were sharing me some of these   decentralized application built on top of 
Arweave. One of them that I was looking at,   was like a Google drive wrapper, like an app that 
looks at Google drive where you can upload like   private documents and store them permanently.
ArDrive.

Yeah.  ArDrive. Yeah, that's the one. Can you 
store private information, I dunno,   encrypted private information on Arweave? Is that 
a good idea or that's kind of a pretty bad idea   to store like private information on Arweave?
This is an interesting question. There's a lot   of different schools of thought. Like obviously 
you should not be storing private information   publicly permanently on the Arweave. That's like 
saying, you know, look before crossing the road   but whether you should store Ciphertext on 
the Arweave, my personal opinion is if you   look carefully at the way, it is architected, 
pretty much anyone can get access to your lines   of communication.

Right. And that's sending huge 
amounts of Ciphertext, and then you're [inaudible]   all the time. So really the difference between 
that and storing the information on ArDrive in   a very heavily encrypted way, not actually that 
high. Yeah, that's what I would say about that.   I think that it's complicated and people are going 
to have different risk models using some kind of   quantum resistant storage encryption system for 
your stored data makes sense.

But that only pushes   the question back further, basically Yeah. So 
I think that comes down to people's individual   risk tolerance as to what makes sense for them.
Yeah. I think at the end of the day, it's that,   how strong is your encryption, like 
what will be the future quantum,   like you say quantum computing can it encrypt, 
can it decrypt whatever, can it brute force any   of the current encryption technology in the 
future. And essentially if you use ArDrive,   you are trusting ArDrive with the encryption. 
What if our driver is malicious, for example?  I mean this is one of the interesting things 
about profit sharing communities in decentralized   Permaweb apps is that basically they are permanent 
by design for the utility that they offer.   It has these quite interesting effects, like 
you can audit the code of ArDrive once and   have it associated with the TXID so it sent to the 
network.

And then if you go back to that TXID and   that TXID has been audited and you're comfortable 
that it is safe and is doing what it's purported   to do, then you can be confident that that 
service will never change. This is kind of vaguely   useful in the case of ArDrive. But really very 
significantly useful in the case of for example,   a email system, an example I always use. So when 
you sign up for an email account with Gmail,   you basically sign over your ability to move away 
by having your identity [inaudible] what you give   to other people. So your email address you give 
to other people, now you're stuck with Google.   And they can change the service they're offering 
to you all the time. Same with Facebook and so on.   Like once you get locked into the service, then 
the company can basically do whatever they want.   And you've got to remember that these companies 
aren't necessarily incentivized by the things   that make the customers happy or successful.
This is really, really dangerous.

I would   even argue that it's like the hidden danger with 
the Web 2 world is that basically the incentive   structure is that you track people into using a 
new platform and people bankroll this process.   And then once they're all in the new 
platform, they can't really easily leave.   Then you turn on the money tax, one way or 
another. And this normally means that the quality   of the service rendered to the users, the clients 
yeah, and people ended up being somewhat exploited   by the system. Whereas on the Permaweb we can 
do something much, much different. We can say   the relationship between the developer and the 
user is fundamentally, the developer can release   new versions of the software, but those new 
versions must increase. There must be strictly   better than the prior versions. Otherwise people 
just don't update and they continue to [inaudible]   software. So the trust model and the relationship 
that was just fundamentally different.   I think that's actually really powerful.
Let's jump into the profit sharing communities   and I understand that some of these profit sharing 
communities, they issued of profit sharing tokens   to the members.

So I guess my first question, 
right, are the profit sharing communities   similar to decentralized autonomous organization, 
DAO? They sound the same to me conceptually,   but I wanted to hear from you. And 
then after that, maybe can you share   some examples of interesting profit sharing 
communities that are in the Arweave community?  Yeah, absolutely. Okay. Are they the same as 
DAO? They are like DAO plus profit sharing   basically. In fact, I used to call them, before 
we did some branding on it, I used to call them   psDAOs, P-S-D-A-O. Because basically the 
same principle, you have a decentralized   autonomous organization, but the critical 
part is that that autonomous organization,   the members are, [inaudible] giving reward direct 
[inaudible] from just holding the tokens when   people use the application. And so now the profit 
sharing tokens associated with the application   become a, you know, API yielding asset for them. 
And they can just see, like, if I hold this for   a year, how much relative to my investment, am 
I going to get back as essentially interest?   Yeah.

So the profit sharing in nature 
rather than accruing value to the treasury.   Now what are some interesting ones in the system? 
There's so many. But I would say for example,   hard drive is a great example, to be honest. 
It's just a permanent decentralized file sync   application. You can easily replace Dropbox and 
Google Drive for most people. And if you own the   token, then you take part of a 20% tip that's 
added when people upload data. Very simple,   very effective, generates a lot 
of revenue. And take another   somewhat, what would you say, their 
self-referential because [inaudible]   itself a profit sharing token exchange.
One of the things it does is it allows people   to issue images as NFTs or what we're essentially 
calling atomic NFTs.

So, this is a little bit   novel and interesting, but basically the 
idea is that with Arweave, because you can   start smart contracts from the tags associated, 
where they file. Then now you can associate the   contract of an NFT directly with the file 
of NFT in one thing, and then refers to   them by the same transaction ID. So they 
have, for example, an atomic NFT exchange,   basically. That's a little bit like OpenSea 
mixed with Binance, somewhere in the middle.   Or you know, OpenSea and Uniswap, perhaps. And 
if you own the [inaudible] profit sharing token,   then you get a small tip every time someone takes 
the transaction.

Or it makes a transaction. Now,   suddenly you financialized in a very real and, 
literally make real time financialization of the   profit created by web applications, which we think 
is really powerful. Now as people just buy into   the profit streams created by the applications 
that they like, when they do so by supporting   the things that they can [inaudible], they get 
a portion of the profit in the future. And the   founders the team get another way of financing 
the development of the thing they're building.  And what about the Arweave token, like what's 
kind of the tokenomics behind the AR token?  Yeah. Well, I mean, it's pretty simple in 
some way. You use it to pay for storage,   but when you pay for storage, instead of the 
token going directly to a miner, like you would in   Bitcoin or Ethereum, Or actually 
before the new Ethereum upgrade,   but it goes into the endowment and then the 
endowment, essentially, [inaudible] release from   the endowment, when the block reward doesn't cover 
the cost of paying for storage in the network   or on its own.

So, essentially right now you're 
not taking any tokens from the endowment because   the profit isn't high enough. And that gives 
us time to basically bootstrap that endowment.   And then over time you just take tokens from the 
endowment relative to the amount required to pay   for storage, plus some profits for the miners.
So the miners, I mean, for Bitcoin miners, they   run ASIC machines, for Ethereum miners, they use 
GPU machines.

What about Arweave miners? Do they   run like massive data warehouses with terabytes of 
storage, or kind of miners are there for Arweave?  Yeah, certainly terabytes of storage. So it's a 
mixture of compute and execute. You store some and   you also do some computation. It's actually what 
we call a proof of work sandwich, this called.   Technically it's for us, to people outside of 
the deep tech world, we just called it a proof of   work sandwich.

What's happening is like you are 
doing a piece of computation and the output of   the computation is a random place in the network 
that you have to prove that you have access to.   And then you grab that data and you package it 
again and you hash against [inaudible] compute   and you check the output of that, to see if that 
essentially satisfies the difficulty required. So   if that is a winning candidate that you're 
trying to mine. Yeah. And so this is the system.  And then I suppose they difficulty and 
then every block can only be mined by   a single miner, I suppose based on the 
difficulty level like Bitcoin or yeah?  Yeah, it's just like Bitcoin in that way. So, 
I mean, technically you could mine two at the   same height, very occasionally happens, 
but again someone would mine one on block,   not one on top, and then you get a fork of it.
Yeah.

What about the inflation model for Arweave?   Like Bitcoin started with 50 Bitcoin per 
block every 10 minutes. What about Arweave   and how does it taper off to asymptote, 
for example, like a max supply of AR token?  Yeah. So the maximum supply is like 66 million.   There's a few million left to mine, something 
like that. It's, fairly low inflation system.  66 million and about 43 million has already been 
mined. So maybe roughly three quarters have been   mined. And we're left with the remaining one 
quarter that we mine in the next, I don't   know how many more years? I'm just curious. Yeah.
It doesn't stop. We use an exponentially decaying   curve. So we actually recalculate the block reward 
per block rather than halving. We actually think   halving is like a super negative component to 
having a network like this. It just makes a   lot of uncertainty in the markets, like what will 
happen after the halving? Who knows? And you see   it in Bitcoin every time people start to panic. 
I mean, you could argue maybe it's good also   because it brings some attention to the protocol. 
Like I dunno.

But we try and keep things stable,   and predictable, and unsurprising. And so, yeah, 
we just have this like decaying curve per block.  Yeah. Because like the halving is such 
a big change in reward for the miners.   It's like first you got like 50, and then 25, it's 
basically half. Like there's always a question   what's going to happen to all this hashing 
power. Will miners stay on or it's no longer   profitable and they will remove everything. 
And then difficulty comes in and then like,   is the networks still secure once the hashing 
power goes down.

And then miners only mine if the   price goes up. So the price sort of have to double 
every four years to sort of make sense, otherwise   there's so much economics and numbers behind 
it. Yeah. I see where you're coming from.  Yeah. It's very interesting component. I 
don't think Satoshi did it deliberately,   is my guess. I think it's just 
like a, well, the math is easier.   No, which is no offense. Like you don't 
want to screw something like this up.   And I know my impression from the Bitcoin 
white paper which was an amazing piece of work,   technically. Not diminishing that. It spawned 
an entire industry that we now will work in,   which is worth like $2 trillion right now. 
Unbelievably cool. But my impression with   the paper was, it's almost unfinished.

Like if you 
really get a feeling when you're reading through   it and he gets to the last page or two, it tapers 
off and like, I dunno, like "I've got something   else to do. I'm going to write this later". And 
then he just stops. It's really quite something.  And you get that, what it's worth with the 
Bitcoin code base is also pretty similarly, like,   with lots of sensible things, but they 
didn't have a block size limit at all,   for example, and no DPoS protection until 
like a week before Satoshi disappeared.   It was an amazing system, but it's like 
sort of cobbled together in some sense.  So I guess one of my last questions, I suppose. 
would be what's kind of the plan for Arweave   this year, next year and next few years? Yeah.
Yeah.

I mean, what's really exciting about the   ecosystem at the moment is it's just reached this 
very clear inflection point where the founders of   protocols, some of them, but startups in general, 
profit sharing communities on top of the network,   who've taken the reins of like developing the 
things on top. And really our job is as the   core Arweave team is just become like supporting 
these people as much as we can. So almost all of   our effort now goes into founders support, 
for people building on top of the platform.   And it's working really, really well.

Like 
amazingly well, it's so cool to see people,   they quit their jobs and they raise capital and 
they go and build the opponent decentralized   web together. It's fantastic to see. Yeah, 
so we're just focused on supporting that.   Everything the ecosystem needs to 
make that happen more effectively.   Even more if I had to be, frankly. Yeah. That's 
the focus in terms of the core protocol itself,   it's pretty much complete. I don't want 
to change too much about it at this point.  There was some things we would like to iron 
out.

We'll keep developing little bits, but   we're very much of the mind that good protocols 
mature and then stop changing. They get to the   point where they will be scalable. They 
effectively solve a problem. And then they   mature and really like freeze. You'll note that 
all of the important protocols in our daily life   basically don't change that often at all. 
Like power sockets, for example, same format,   dozens of years. Yeah. And there's actually 
an interesting reason. This is very important,   which is basically people need to trust 
that the protocol isn't going to change   In order to embed it in things and the more 
trusted it is, the more I can embed it in   stuff. You imagine, it's an interesting code 
experiment, if the USB standard hadn't changed   like five times in the last, you know, 20 years, 
I guess, how much sooner would we have seen USB   ports embedded into a power sockets that are 
now found.

I think almost instantly, actually   It's an obviously useful thing to have, like 
you've been carrying around these tiny adapters   for many, many years. And that I think is because 
people couldn't trust the protocol not to change   under their feet. And so there's a very 
important thing that needs to happen.   You see it with Bitcoin now. Basically it's 
impossible to change Bitcoin in any serious way.   Because we've all bought into, this is the 
archetype of what Bitcoin is until always be.   Yeah, I think that process needs to happen 
and it is happening.

That means that our   attention now is on the layers on top of the 
base protocol. Things like SmartWeave, the   smart contracting platform allows you to run 
arbitrary amounts of compute without gas fees   with actually just storage fees. Yeah, 
all sorts of stuff like this and what not.  Interesting perspective there you brought up 
about how protocols shouldn't change. Seems to be,   I mean, Bitcoin is, you know, it's 
impossible to change, but it seems to be   quite a different perspective with Ethereum, 
where there seems to be constant changing of   the protocols with the recent EIP and the 
miners are getting angry right now. And   there's ETH 2 massive, massive changes 
coming along as well. So quite different   philosophy of how things are going about.
Well, and I think part of that is because   everybody in the heads knows that a huge change 
has to happen to Ethereum for Ethereum to survive.   That is ETH 2, basically. And so we've got in 
a heads, yeah, you can change whatever you want   about the protocol. We're not going to buy into it 
that deeply because we know it's going to shift.   And we've been thinking it will shift 
in a year, for the last three years.  And the thing about it is, ETH 2, I mean, you 
brought up a point, like, my mindset is set   that I mean, Bitcoin's never going to change.

If 
it change, someone's gonna fork it and go back   to its origin idea. Whereas ETH is just going 
to be constantly changing. There's ETH 2, but   there'll potentially be ETH 3, ETH 4, ETH 5 and 
they will constantly evolve, where it's kind of   like the USB port analogy, like, I won't even be 
surprised if ETH keeps evolving over the years.  Yeah, but I think if you do that it's not going 
to be able to really decentralize properly and   people won't trust it in the same way that they do 
Bitcoin. Bitcoin has a shot at being digital gold,   if not just digital currency, full 
stop, because people can trust it.   We think that archive are essentially 
the same thing. This idea of like, "Hey,   we give you the opportunity to store your 
blogposts in perpetuity, literally forever,   replicated all over the world, no one can alter 
it, no one can take it away. No one censor it   at any point in the future". We offer 
that for like a fraction of a cent.   And I think that we're going to need to come 
to like a conclusion on what the protocol is   in order for people to be able to trust that.

In 
the same way, to loop it back to the beginning,   we were talking about, you know, this 0.5% or 
200 years thing, you need to give it some kind of   really robust conservativeness in order for people 
to adopt it and use it for its intended purpose.   Yeah. It's interesting discussion for sure.
Yeah. Thanks a lot for taking the time to   explain Arweave to us.

Definitely very 
insightful. I learned a few things over   here and I'm sure anyone listening to this 
will also learn a few things about Arweave.  Yeah. Thanks so much for the 
conversation. It's great.  Thank you..

You May Also Like