The Limits of Blockchain Tech

28 Nov 2014

Putting software "on a blockchain" is an inconvenience, justified only in cases involving a global public ledger. These ledger-cases are likely to be limited to Money, Reputation, and Names. This limitation suggests that other "DACs" are not economically viable, and Ethereum may be "a solution in search of a problem".

Are there really 84 different blockchain-uses? Will "a thousand chains blossom? If Bitcoin is just the "first DAC", then how many DACs will there be? Will there be many distinct sidechains, or will they be mostly for staging upgrades? How many Ethereum contracts will we see in use?

Why would we use blockchains at all?

What Blockchains Do

Blockchain = CryptoSignatures + PoW Clock = "Who owns what, when?"

The Blockchain Niche

Blockchains can only provide codeable services (digital information only, nothing physical). This limitation aside, Blockchains are very reliable (software always does the same job, regardless of the time of day, location, or the user’s age/race/religion/criminal-record/intentions).

Bitcoin is a perfect example: wealth transfers were easy to code (basic arithmetic), and the marginal reliability was extreme ({taxes, inflation, bank hours, e-gold} were unreliable in the sense that you were treated worse if you were {high-income, unbanked/unpopular, day-job-employed, e-gold customer}).

Where Reliability Is Crucial: Owning Assets

Compare the bankruptcy of two different types of business: [1] "money-accepters" (like retailers), and [2] "money-storers" (like banks).

Money-accepters can go bankrupt almost painlessly: Soon a new one will appear to fill the unmet market demand, and you can continue your shopping there (think of Bitcoin’s "Silk Road(s)"). If you are extremely unlucky, and happen to be shopping mid-bankruptcy, you might permanently lose some value: that of the single transaction you were mid-conducting.

However, if a bank (money-storer) closes, an entire lifetime of savings can be wiped out permanently. If a new bank opened (from where does it get its initial deposits?), users would have to start saving all over again. Money-storage has higher stakes: owners/employees/hackers/lawyers/politicians can target the stored asset for theft, extracting a "ledger rent". The damage is permanent, and the pain excruciating.

The story isn’t limited to bankruptcy (it is simply easier to use this extreme to see the difference between accepters and storers). Instead, imagine each type did anything you don’t like. With money-accepters (retailers) you can take your business elsewhere. For money-storers, if you’ve already deposited you are screwed. All you ever owned was a promise to be repaid: the usefulness of that promise is defined by its reliability.

Storage and Retrieval

The "ownership" concept implies certain requirements (one can’t "own" the fresh, outdoor air). Blockchain tech (an uncensorable database and a distributed timestamp server) enabled digital assets to meet these requirements. However, the assets themselves are still digital: while it may one day be possible to buy a "smart property" car using only the Bitcoin blockchain, the value of the car will not be protected from a sledgehammer in the same way that each Bitcoin is protected (from a sledgehammer) by the blockchain.

Ledgers: Dimensions of Programmable Value

Normally, "what other people own" doesn’t affect "what you own". So far, the rare exceptions are: Money, Status, and Names.

Money / Finance (Bitcoin)

Raw value-storage? Bitcoin. Stocks/Bonds/"Wall St."? Colored Bitcoins. And more besides: we have multisig, nLockTime, and sequence numbers, which greatly enhance the user’s security and transfer experience. When Andreas Antonopoulos says that "currency is just the first app", he means apps involving the Bitcoin protocol, not new blockchain-apps/protocols.

Bitcoin is only weak on purposefully storing value (it only describes where value is currently stored). Those businesses which store customer’s Bitcoin (namely exchanges and betting sites) routinely lose or steal those funds. Bitcoin is P2P money: it was not designed for someone else to control your money. It was designed for you to control your money.

Status / Reputation / Expertise / Branding (Truthcoin’s "Votecoins")

Purposeful value-storage is what I created Truthcoin to address: it has blockchain rules which store up Bitcoin and conditionally pay it out…a prediction marketplace containing a trustless outcome-resolution service (ie, an escrow running entirely on greed and not on second-order abstractions such as ‘honesty’).

To do this, Truthcoin had to add a new dimension of ownership to the blockchain: that of ‘reputation’ (the Votecoins). Some individuals may have greater subject-matter expertise than others: when groups disagree it is best to weight individual opinions according to their relative expertise. We must divide our limited attention span, and consider only the most-relevant knowledge-sources. Correspondingly, status is a zero-sum game: if you’re on top, you lose when others do well, and if you’re below, you benefit from the misfortune of others.

Identities / Names (Namecoin)

I searched for other zero-sum, ‘necessarily-scarce assets’, but I could only find one other: Names (you can’t have someone taking yours, for example). The case for digially-scarce names already been made, but, more importantly, the name-software itself has already been developed!

The DAC Delusions

Some people want to "decentralize all the things", not realizing that most of them are already decentralized.

Many have made the following mistakes:

#1 Forgetting that Our Economy Is Already Decentralized

With Bitcoin and an internet connection, anyone can start a business selling anything. This takes an already-decentralized economy, and turns the decentralization up to 11. Some services require customers to escrow their Bitcoin, which may require the creative use of multisig, reputation, insurance, fidelity bonds, etc. Finally, there are business arrangements which are impossible without a trustless escrow. These (few) services can use Truthcoin (and tap the reputation dimension of ownership).

Unless you’ve come across another "dimension of digital scarcity" (see above), blockchains aren’t worth your effort! Just start a normal business instead! You can flexibly adjust your prices and service offerings as you, the manager and entrepreneur, see fit! .

(Note that there have been many SilkRoad businesses using Bitcoin but no SilkRoad DACs using the blockchain…coding the required functionality would take too much work).

Some things aren’t codeable, and "reliability" can be very constraining (you’ll want to treat some customers differently from others). Software bugs, miner attacks…these can crash blockchains permanently! Put those skills toward writing a great piece of software (which you sell for Bitcoin)

Which brings me to…

#2 Confusing "DACs" with "Software"

When I claim that that the blockchain is an inefficient model, I am not claiming that software is an inefficient model.

Our world contains several "software businesses". Google / Facebook run operations in almost the same way that Bitcoin does: a group of programmers meeting to collaborate on software. New software might be sold for Bitcoin, or used to build a business which earns Bitcoin. However, that "the tech will improve" doesn’t change the relative disadvantages a blockchain has over a firm.

Take Storj, for example. With the tech tools we already have, anyone can already sell or rent their spare hard-drive space. Obviously, new software can help coordinate buyers, sellers, payments, and terms, but this software does not have to use a blockchain, it only needs to use Bitcoin the currency.

The freedom to manage can be very helpful, but a business can remove these freedoms as required (consider the zero-knowledge backup service SpiderOak, who’s employees cannot read user’s files). The same goes for transparency (think provably-fair online gambling sites), and even for decentralization (think Bitmessage, or Bittorrent, which are completely decentralized but have no need for a distributed clock). These are purely software/tech improvements. No blockchain required.

#3 Ignoring Software Rot

Software development and maintenance requires a great deal of highly-skilled work. On top of that, post-development you still have all your work ahead of you, because software "rots" as it gradually becomes obsolete. Bitcoin is supported by users in forums, in lieu of salesmen/customer-service-reps. The work is done by volunteers, but it still requires effort. Generally, open source software projects can thrive on community-autopilot, but they are also regularly abandoned (and many Bitcoin-projects have already achieved vaporware status). Bitcoin’s unique status as money not only incentivizes development/maintenance but also co-opts money’s-status-as-a-network-effect to onboard and coordinate programmers.

It may be difficult for programmers to believe (or accept), but software is long-run inefficient. An entrepreneur can easily copy a DAC, especially an open source one, with a website and some free time. The reverse, a DAC replicating even a simple business model, is a monumental project. Ethereum is a perfect example of this: anyone can start "Ethereum.com", collect and run contracts, and then manually authorize payments to beneficiaries. Blockchain-Ethereum has taken months of highly-skilled effort. Is Blockchain-Ethereum worth the wait? That depends on the codeability-reliability tradeoff (for a start).

But even with Ethereum built and fully-functional, each contract would require extensive development, be adamantly inflexible, and suffer software rot. With all labor likely unpaid…why do this? Just start a website and become an entrepreneur!

Conclusion

The Bitcoin Blockchain was invented 6 years ago. Since then, Bitcoin has been used by thousands of thriving firms, but no DACs yet exist (with the possible exception of the very first alt-chain, Namecoin). In general, the Blockchain has only been used for one other useful project (Namecoin), and my own project (Truthcoin) might be the third and last. While more Bitcoin-accepters (and even Bitcoin-business) are planned each day, even the Ethereum forum doesn’t have a single concrete plan for an on-Blockchain business ("DAC") (in fact, I could not find any concrete plan for any final good at all…point them out in the comments if you find them).

Although this essay concerns the "why" of blockchains, it does not concern the "how". New blockchain-engineering can certainly provide a better way of managing money, status, and names. On that we will likely still see tremendous innovation.

Update: A previous version of this essay included Bitmessage as a "useful blockchain project", until a reader pointed out via tweet that Bitmessage does not use a blockchain. (Which makes perfect sense, as "other people’s messages" don’t affect "your messages"). I was allowing Namecoin and Bitmessage to share the ‘Name’ concept, even though BM is actually designed to use NMC for identities. This version has been edited to clarify that Namecoin is the useful blockchain project, not Bitmessage.

Also on Truthcoin

Measuring Decentralization

  • 5 months ago
Bitcoin’s Decentralization increases as a full node becomes cheaper. …

Nothing is Cheaper than Proof of Work

  • 6 months ago
So-called "alternatives" to Proof-of-Work "waste" just as much "work". Intro: …

Expensive Privacy is Useless Privacy

  • 9 months ago
Privacy is impersonation. To be useful, a privacy tech must be [1] cheaper, or at …

Long Live Proof-of-Work, Long Live …

  • 7 months ago
Satoshi got it right: [1] proof of work ensures distributed consensus, and [2] …

Security Budget in the Long Run

  • 9 months ago
A discussion of Bitcoin’s ability to resist 51% attacks (ie, its "security budget"). …

Truthcoin: Making Cheap Talk Expensive

  • 9 months ago
Has something happened to "Bitcoin Maximalism"? Is Bitcoin "Qwerty" …

Nothing is Cheaper than Proof of Work

  • 8 months ago
So-called "alternatives" to Proof-of-Work "waste" just as much "work". Intro: …
× Comments for this thread are now closed
  • Vitalik Buterin • 7 years ago • edited

    And I disagree as usual :) Reputation is actually the one oft-cited blockchain use case where blockchains are _not_ particularly useful. The reason is simple: status is not tradeable. Each individual's status/reputation is simply the sum of other people's opinions about them, and people have the ability to change their opinions at any time. The whole point of money is that once you spend it you can't take it back. The whole point of reputation is that when you assign someone reputation you _can_ take it back. If you want to know someone's reputation, you just ask other people/bots what they think about them via network messages. The only reason we care about _on-blockchain reputation_ for Ethereum is so that good old blockchain-based DACs can use the reputation securely.

    You tried to make an argument in favor of Truthcoin, and instead you ended up advocating Codius.

    So what's the purpose of blockchains? Simple: blockchains are useful for decentralized consensus on databases that update themselves according to non-commutative (ie. order-dependent) state transition functions. Currency and names are two use cases. Intermediate state of business relationships is a third. Timestamping events is a fourth. There is an exponentially infinite number of non-commutative state transition functions; at this point I'm not going to agree with anyone that claims anything close to an exhaustive categorization.

      • In a world with IT/crypto upgrades, and changing passwords (etc), it seems that reputation must be transferable across a database, and in a pseudo-private world (with only keypairs, like Bitcoin), this would imply tradeablity. The point is not to make something which *resembles the appearance* of real-world reputation, but something which *replicates the function* of real-world reputation (something that certain groups want to build up and maintain over time).

        I've never heard of Codius until now, but (from what I've been reading) it seems to fall victim directly to #3.

        I think you've over-simplified your "purpose", and it is instead a "definition". Where do we need a "decentralized consensus on databases" (your purpose)? As I point out, the contents of the hard drive of *someone else* usually don't affect you, nor do shoddy business practices at places where *other people* shop. The Satoshian PoW-Clock can provide secure time-stamping (only while the miners get paid in something they can use), but there are many ways of timestamping.

          • > The point is not to make something which *resembles the appearance* of real-world reputation, but something which *replicates the function* of real-world reputation (something that certain groups want to build up and maintain over time).

            Reputation has two purposes. First, it's a sort of poor man's security deposit: McDonalds knows to heavily police their food safety practices because if they don't and they have an incident, then they'll get a big takedown in the media and lose a significant portion of customers overnight. Knowing this, consumers are more likely to shop at places that have good reputations, and knowing this stores strongly desire to keep their reputations - one of those funny self-enforcing metastable equilibria. Second, reputation is a predictor of personality traits (or in an institution's case, governance practices) and hence behavior. Intra-protocol tradeable reputation keeps the first function but loses the second. However, if you just care about the first function, then in the cryptocurrency day and age it may make more sense to just use a plain old security deposit - which is what truthcoin reputation actually is.

            > The point is not to make something which *resembles the appearance* of real-world reputation, but something which *replicates the function*

            Not sure what the appearance/function distinction that you are trying to make is.

              • Reputation is far more than a poor man's security deposit. Debt and Equity are very different. Equity is far more flexible than debt, and in times of crises, debt-to-equity-conversions are often the best saving throw (for everyone involved). In fact, in a few ways, debt IS equity because the repayments are a function of the borrower's competence, so a security deposit is more like a poor man's reputation.

            • Your silk road caption is right on the money, there's no Silk Road DAC because "coding the required functionality would take too much work". Take OpenBazaar, they are trying to use bitcoin for money, namecoin for names, and rolling their own quasi-DHT for reputation. Wouldn't it make sense to use one general-purpose blockchain for all three (money, names, and reputation)?

              OpenBazaar is making progress, but its only one DApp. Other DApps would have to duplicate their effort, or be forks of OpenBazaar. Its that kind of forking (one-off duplicate codebases) which leads to code rot. What's needed is a platform for DApps, which would make it easy to create, publish, and use a wide variety of "serverless" services (ie. p2p architecture). Creating such a platform requires a lot of work (I'd even say tremendous innovation ;). And yes, it will be a lot of software, but its worth trying because of the distinct advantages a p2p architecture offers over "normal business" centrally-operated websites.

                • What about BitShares and the family? Do you know what happens inside NXT? What about Nubits?
                  What about all DApps projects being collected by cyber•Talks? http://cybertalks.org/c/dapps. It seems that you feel yourself the only guy who understand the stuff. Frankly we don't see any truth at all in your project. At least for now. And that's unclear there your truth should come from.

                    • I actually circulated an earlier version of this essay for feedback on the Bitshares forum. It was a very productive conversation but ultimately I think my ideas survived the gauntlet. https://bitsharestalk.org/i...

                      As to NXT, I'm afraid the 100% premine makes NXT a non-starter, and the fact that its marketcap is below that of the statirical DogeCoin proves that there's no reason to spend any valuable time discussing it.

                      Nubits has good intentions but is flawed...that happens to be my next blog post, so stay tuned. Unfortunately I work full time and can only work on blog stuff about an hour per day.

                      re: cybertalks apps. What about them? My guess is still that, overwhelmingly (maybe not always), a business will be able to provide more value to the user than a blockchain. For the youtube example, the type of user who watches video entertainment usually finds it more psychologically comfortable to watch a video ad than make an upfront monetary payment. "Watchmybit" is focused on producers, not users, which is an obvious mistake. Youtube gets the lion's share of the ad-proceeds because it had to build and maintain the infrastructure which delivers the **user** experience (including, most importantly, the brand, and also the menu of consistent [fallback] content-options).

                      • Casey Detrio • 7 years ago • edited

                        How about Paul Brody's reasons for choosing a blockchain-based architecture for Internet of Things infrastructure at IBM? (https://gigaom.com/2014/09/... his interview starts at 22:00). Are those honest tech/econ concerns? Or just riding the hype train?

                        Edit: and an update since the podcast interview

                          • I'm ashamed to admit I couldn't make it through the video. So much empty-suit, fluff, buzzword-nonsense. I tried to skip through to find any actual information-content, but was unsuccessful. Could you do me a huge favor and pick your "favorite" 5 minutes or something? From 19:00 he seemed to admit that he had no case.

                            If anything, this presentation seems to be a perfect example of mindless hype. From what I saw, he actually didn't give a single "reason for choosing a blockchain-based architecture". Just kind of described a bunch of things at random. Am I wrong?

                            The first questioner nails it, I think. The presenter then stalls for time while fabricating more nonsense. In fact, the Q&A period makes me feel that my views about this are actually the mainstream ("silent majority"), its just that those who "figure it out" decide to leave the conversation, and don't spend their valuable time going back to explain it to those who are still catching up.

                              • You're not wrong, I too was disappointed by the video. There were some good insights ("we need smarter toasters"), but it often sounded like a markov chain text generator of buzzwords. I thought his answer to the question about redundancy vs scalability was sufficient: all things won't be on one single blockchain, different sets will live in their own "namespaces" (only the devices in one house would be on one chain, or all the devices in one factory, etc).

                                Paul Brody's interview on the Gigaom podcast was worthwhile. He expects conventional cloud platforms to be too expensive for any one device manufacturer to subsidize. Because there are so many "things"-devices, and they last for 10+ years, operating costs for a centralized cloud start to really add up (and shutting down five years down the road would infuriate device owners). Additionally, free/ad-based services are unworkable (device usage data has zero marginal cost).

                                The blockchain is for device-device contracts, and to enable "autonomous distributed marketplaces, not just to share money (which you could do with bitcoin)". He says to do that you need distributed transaction processing, meaning generalized database transactions, not just financial ledger transactions.

                                The use of Telehash (a p2p messaging protocol) complements the blockchain (in his example, after flipping a light switch, you dont want to wait for a block to confirm before the light on). As you point out, Bitmessage (also p2p messaging) is decentralized but it is not a blockchain. Its really the combination of the two (plus a third, p2p filesharing component) that yields an elegant system.

                                  see more

                              Older · View Archive (58)

                              Altcoins Aren't Money, They're Bitcoin's Casino/Laundromat

                              Altcoins aren’t money in theory or practice. Instead, they represent gambling and/or money-laundering opportunities. One result of this is a confused public dialog about cryptocurrency.

                              Newer

                              BitUSD Isn't Worth The Trouble

                              The BitUSD isn’t very useful, should NOT be worth exactly $1 at all times, and (like all assets) requires fundamental value. Truthcoin might indirectly create BitUSD as a result of the USD’s inability to be used in prediction markets.

                              ©2021 Truthcoin