Hacker News new | comments | show | ask | jobs | submit login
The blockchain story is bullshit (geoffreywoo.com)
43 points by geoffwoo on Apr 24, 2014 | hide | past | web | favorite | 27 comments

The author is missing the fact that the world has adapted to the requirement for centralised trust. That makes it very hard to see the value of a blockchain system - you automatically think about problems and solutions within another constrained framework. It's similar to e-mail before people used email. It seems laughable now, but it wasn't obvious why e-mail would be useful ("I need to switch on my computer every time to check if I have a message!? A phone is so much easier..."). The value of a blockchain system will become more and more evident over time as we use it to solve problems that are hard to see in our current paradigm.

Comparing the blockchain directly with centralised switching systems is like comparing e-mail with phone calls, and using "ability to speak" as the metric of comparison. E-mail allows you to do something else, but it took us a while to realise that something else is incredibly valuable. The blockchain will be the same. Simple example: we don't even think in terms of individuals or corporations issuing their own currency. It's not a "problem" that we generally acknowledge to exist. We'll certainly see it as one worth solving when companies and rich individuals start using blockchain technology to compete with countries in terms of issuing currency.

You're making strawman arguments e.g. a blockchain and a centralized database is much more similar than email and phone.

I think your comment is affirming my statement implicitly. For the things you want to do with a centralised database they are very similar. Given the entirety of things you can do with a blockchain then are very different. You view email and phones as very different, because enough time has passed for those differences to become obvious.

No, I think there's clear objective attributes that are different. Email vs. phone: async vs. sync, recorded vs. unrecorded, etc. The only real difference between blockchain vs. standard db is anyone can write who has 'x-coin' vs. trusted authority can write, and I argue that the difference really isn't that important.

I think if you replace phone with 'answering machine', his logic makes more sense.

The article makes three main arguments that have been debated endlessly the last few years.

1) Blockchains are not as fast as centralized databases.

This is true in the same way that a democracy can't act as fast as a dictatorship. Blockchains are important because they allow a large number of strangers to trade with each other, to exchange value, free from centralized control, corruption and skimming. Millisecond response times would be nice, but that's a secondary feature.

2) The first people to mine or invest in a new cryptocurrency can be far wealthier than those who come later.

How is this different or worse than the status quo? Those who invest their time, money and ingenuity in a risky project reap rewards if it succeeds.

3) Centralized authority is required for arbitration and reversing transactions.

This assumption is exactly what blockchains and cryptocurrencies are testing. Can people exchange value freely, without a central authority that skims unearned value for itself? Can we enforce complex agreements without submitting to a power that stifles innovation?

The final chapter hasn't yet been written, but the experiment looks very promising so far.

1. Usually it takes a 10x improvement to make a switch from status quo to something new. To see a switch when the performance is much worse forces the benefits of blockchain to be extraordinarily compelling. And some of the benefits are not as rosy as initially conceived.

2. Agreed. This is status quo. I see a lot of arguments where folks claim that the blockchain creates a more egalitarian system. I wanted to explain why this wasn't necessarily the case here.

3. Agreed. However, the most important transactions are the ones most prone to arbitration, and arbitration requires authorities. Outside of valuable transactions is there really a need for such a complex database?

Thanks for the thoughtful response.

1. One use case for this is trade settlement. Instead of the DTCC's status quo of T+3(days), if there are trades that can be settled on a blockchain, that can be reduced to T+1(hr) or less. Same with payment/transfer - ACH T+3 sucks, blockchain T+1hr or less is much better. A lot of research into both right now.

3. This one is interesting as there are some areas of law that computational law researchers have decided are uncomputable, hence requiring judges, arbitration, etc. Those probably won't be represented blockchains for the foreseeable future, at least until strong AI is perfected.

There's definitely a lot of noise about moving everything onto blockchains, but as you point not all of it makes sense. The name of the game, especially for startups and companies in this space, is to find what does.

Arbitration requires authority, not authorities. Authority can be granted with limited scope, for a limited time, after which the authority vanishes. Authorities eventually must be bribed or defeated in battle due to scope creep.

The author is making valid points.

Nobody here likes to hear it, and I feel like a lot of people in SV are tuning this out, but bitcoin is really, really struggling to find a relevant use case, especially with consumers. Regular consumers have absolutely no reason to use bitcoin. The "1-click" payment and 1% price discount are not appealing enough to the average Joe who already gets 1-2% cash back, airline miles, and consumer protection on his credit card (and 1-click checkouts on many e-commerce platforms). And more regulation isn't helping the "crypto-anarchist" decentralization angle either. Add price volatility to that and it kills the consumer use case.

Bitcoin is great, FOR MERCHANTS.

BUT PAYMENTS IS A TWO-SIDED MARKET. A winning product must appeal to both sides, the buyer and the seller. Bitcoin appeals to the seller at the expense of the buyer. Even if the blockchain technology overcomes all of its flaws, we would have bootstrapped a value exchange system that does not offer greater benefit to the one that exists today.

Many are making the flawed and simplistic analogy of comparing bitcoin/blockchain technology to the early days of the Internet... That analogy is not valid. When it comes to money, consumers actually WANT a central authority. They want to be insured and cuddled and protected and not run the risk (however small) of being criminally liable for transactions and have someone to speak with if they make an erroneous transaction or have their card stolen. And they can already do all that, which makes it tremendously difficult to compel them to change the habits they've had in forever to adopt a system that does not provide them with significant advantages.

I've been in bitcoin since 2011 and used to be a big believer, but I don't see bitcoin or blockchain technology gaining traction with consumers, because it simply doesn't solve a problem for them.

It could be destined to eternally remain a store of value (like gold) or find targeted applications (like machine-to-machine payments). The underlying blockchain technology could be of more use if adopted by the banking system, for example, but that is still far in the future.

There are some systems where blockchain technology will be really useful (#1 for me is zerocash for non-government-sanctioned transactions, but that probably won't be ready for real use until 2015; not the least of which because we need more secure wallets and other systems to handle a fully anonymous irrevocable currency than even what we have for bitcoin today).

I don't think it's going to take over everything. One, a decentralized network of distinct currencies with an exchange is a better route to decentralization than centrally-coordinated protocol-decentralized currencies for a variety of reasons (performance, ease of modeling, ...); two, a lot of things are inherently centralized.

On OP's third point about reversibility being a feature -- the solution to that is to leave funds in escrow. So, you can do $100 transactions as long as you have $100 at-risk in an escrow transaction, and can do an aggregate volume of transactions up to the amount you have on deposit. Maybe some more sophisticated risk modeling there, too (older accounts, lower risk transactions, etc. take less of your deposit balance).

For escrow, are we then not transferring trust from just one arbitrator to another now? Why re-invent the existing escrow system :)

multi-sig on the blockchain allows you to do escrow in some interesting ways.

> So the trust issue reverts into trusting that the creators + early adopters to not abuse the system. This is the exact argument used against single-authority systems that the authority may abuse the system.

This is not exactly true. Look at Napster vs BitTorrent. Napster failed because it was a single-authority system. BitTorrent has succeeded because it is (mostly) decentralized. Trackers might go the way of the block chain in the not-too-distant future.

As far as abuse: It is up to early adopters to sniff out the scams. Unfortunately that is how free markets work. If those adopters stay and continue to promote the new system, it may have a chance with the populace at large. The small localization of failure is a feature not a bug.

I'm not trying to argue that decentralization is bad (Napster vs. BitTorrent is a great example). And there's a lot of cool technologies around decentralized / anonymized networking. My central point is that a trustless, distributed database i.e. the blockchain doesn't really offer practical advantages / new applications over what can be done today w/o such a database.

You are correct that "functionally" there may be no advantage. The value is not the functionality but the architecture. It is resilient to some very powerful attackers, both actual and theoretical. _That_ is where it derives it's power, not necessarily from functionality.

If Bitcoin operated on a centralized database, as E-Gold did, it would have gotten shut down quickly (as E-Gold did).


Pre-mining doesn't necessarily translate into more power, just more fruits thereof. Pre-mining bitcoin wouldn't mean that one can control the blockchain; it just makes one rich. One might even see that as the just reward of the creator of the technology

Interestingly, in a proof-of-burn system pre-mining would give the pre-miner increased power.

Yeah, I make the distinction that while the distribution of power is relatively fair (determined strictly by processing power), the distribution of wealth (in the form of x-coins) is heavily skewed.

I don't have a problem with early adopters getting their just fruits, but to claim that it's more egalitarian than what exists today is misleading. The same argument can be claimed that Godaddy and other early DNS registrars are getting their just fruits by being early adopters of the current domain name registering system.

Pre-mining and fixed generation rates aren't inherent to blockchain technology, so I think that distracts from your point.

I wanted to note how it's used in practice.

Your 'Less Performant' paragraph seems to misunderstand where the big CPU hit happens: It happens with database inserts (which require mining a valid nonce), not with verifies (which only require verifying that a transaction block has a specified hash). So reads/verifies are fast, but inserts/updates/deletes are slow. Also, the mining parameters can be tuned to make it easier/faster to do an insert, at a tradeoff of a larger blockchain history for initial clients to process and more orphans (See litecoin, which has a 2.5min block processing target). So I think you should consider that a 'tuning parameter' based on your application, more than a set-in-stone disadvantage.

Human administrators are not a feature. Chargebacks and other forms of manual intervention allow a third party to diddle with the business of two transacting parties on the instruction of a single one of them and some idea/law/regulation about what is 'right'. To assume that this is a good idea is hopelessly juvenile and naive.

Try to UML-ify a system with manual intervention. Now modify it accepting the ideas that a) people are imperfect; and b) people work on their own incentives which may not (probably won't) be aligned with the users of the system. The manual intervention bit is now a black hole of undefined behaviour with nothing but a wing and a prayer to say it might actually do something useful.

Try and get that past QA.

Charging back is protection for either one of the two transacting parties to protect themselves from misrepresentation from each other. The fact that it's one of the most popular features of a credit card would imply placing the naiveness on you. Laws/regulation exist because people are imperfect!

Equating popularity with validity is a logical fallacy. Sticking my face in a steaming pile of human shit would be the most popular feature of forced captivity if my captors offered only one other option of molten lead.

Laws and regulations exist because there is an 'elite' who make them, they affect us because the elite have an army of muscle men called the police who threaten our liberty for non-compliance. This is easily, formally provable. The intentions of the elite are murky and decidedly unprovable - how many NSAs do we need to create before people get that?

Regarding performance/cost, there's no mention of Proof of Stake-based blockchains or variants (http://bitshares.org/security/delegated-proof-of-stake.php)

This is an interesting piece. But I think a problem with it is that it assumes an initial unequal distribution of coins is untenable. Has it been a problem for gold that the initial distribution is heavily weighted towards gold miners and their investors?

Addendum (4/26/2014): My focus is from a Computer Systems and Engineering perspective on the blockchain as a distributed database with trustless write access and it’s potential as a major component for new applications. Bitcoin, the application it was designed for, isn’t the focus here.

From a technical standpoint, the blockchain doesn’t open up any new ‘technical’ attributes. The blockchain does not enable any new sensors, nor does it enable an order of magnitude increase in bandwidth, storage, affordability, that doesn’t also apply to any other types of distributed stores of data. In fact, performance will be most probably be worse and at best will not be any better than existing trustful distributed datastores. In other words, any system built on a blockchain can be built on any other distributed datastore.

It does open up new ‘social’ attributes where trust is distributed away from existing central authorities, and there’s a line of argument that existing central authorities slow down innovation in regulated areas. We already see activity around finance and a new form of DNS. My point here is that trust is not magically removed from the story. Trust (in an economic sense) is simply transferred from an existing central authority to a new oligarchy of coredevs and early adopters.

Authority is a different sense, and in the cleanest arguments for blockchain’s potential, authority is transferred from humans to mathematics (any blockchain-based system transferring power to another set of humans is just not that interesting). My line here is that more important transactions tend to be ones that are disputed and thus regulated. Enforcing these decisions (especially in cases of real property and physical assets) therefore tend to be a function of the government. In one scenario if we follow this line to completion, I imagine a computer authority with drones and robots enforcing all contracts. This is perhaps a more trusted government because the blockchain-based datastores will be ground truth and there will be no need for human intervention. I don’t have any qualms against such a future. In this case, authority and trust is transferred from existing governments to properly carry out the will of the people (which a lot of people at any given time don’t think is happening), to authority and trust that every component of this humanless contract enforcement apparatus being properly implemented. But in the immediate and foreseeable timeframe, it’s unclear how this new authority will play with existing authority, and that’s really out of scope for this particular commentary.

Thanks for the responses and discussion!

Guidelines | FAQ | Support | API | Security | Lists | Bookmarklet | Legal | Apply to YC | Contact