Hacker News new | comments | show | ask | jobs | submit login
Explanation of DAICOs (ethresear.ch)
146 points by sethbannon 11 months ago | hide | past | web | favorite | 25 comments

So that 51% attack seems pretty critical if you don't know who owns the coins:

"Dishonest" developers could start with 100% of coin and then sell 49% of coins. As long as the developers retain 51% they can control the entire voting process.

This makes me think you should have a clear accounting of real people to each coin for something like this to work. Otherwise you're vulnerable to devs using shells to scam or bribing holders to increase tap. You're better off cutting out the middle man and just sending your coins to the "honest" developer directly because you really don't have any recourse when that happens.

Vitalik's assumption that developers will always be "honest" seems kind of silly because if developers are always honest what's the point of this complicated system he's set up? Might as well just give funds away to our "honest" developers' address.

In the boring fusty old world of law, where the past lives but still somehow operates pretty well, there are two concepts that render this tractable.

The first is process serving. To start a lawsuit you need to be able to identify and notify the other party (except in very special circumstances). If you transact with anonymous counterparties, you won't get much help from the courts.

The second is a nexus. Basically, it's something the court has the authority to reach out and grab. It might be property, a passport, a right, money, accounts, whatever. So long as the court can make an order and someone, somewhere will enforce it, there's a nexus. Without a nexus a court's order is just warm paper. Wave it all you like, the other guy won't feel any heat.

If you don't have a nexus -- if trust and authority is totally distributed, relying on code you hope is the first bug-free code you've ever written in your life -- then it's pretty hard to go from starting a dispute to receiving compensation or enforcing retribution.

Distributing trust sounds really cool, but it turns out that the unit economics suck. There are economies of scale involved for both malicious actors and those who want to punish them. In the current ecosystem, the adversaries are few and their interests highly concentrated. The defenders are many, widely dispersed and poorly able to coordinate their efforts. This is the sort of situation centralised trust is much better at sorting out.

I am of course not a lawyer and nothing here is not legal advice.

I should add that there's one area where blockchain systems have a pretty strong advantage: establishing a sequence of events.

Court cases often come down to who had the better evidence. If you sue me saying you never approved something, but I have meeting minutes showing otherwise, then I am more less going to win by default. Likewise if I say you never had authority to withdraw money from the company account, but you can produce a copy of an authority letter held by the bank with your name on it, you win.

Blockchain promises to shortcut a lot of evidentiary stuff, which is a win. That's why there are so many banks interested in distributed ledgers, title registries, supply chain custody and so on.

All that a blockchain does is establish a sequence of events. It doesn't have to be a sequence of state transitions, it just usually has been.

I don't really follow what you're saying, but I assume you're pointing out an implication I've missed.

Well, the developers would still need to actually put in enough money to have 51% of the tokens, right? The rules for getting tokens are open, and presumably ones that simply grant tokens to the developers will be rightfully ignored.

(This is assuming the ICO investors become more sane in the first place; otherwise DAICOs are no use anyway)

If you’re buying from yourself, things are very cheap- spend the money you just paid yourself for the first batch of coin to buy the second, repeat as required to own 51%.

Perhaps thus could be partially circumvented by requiring the sale of all coins to be simultaneous, but a reasonably clever person could use other financial machinery (loans, etc) to avhieve the same effect.

But the point is you don't get the money you spent on the tokens. It's locked into the contract - all you can withdraw is the eth/s defined in the contract, and that only after the purchase period is closed.

Ah, you're right- I shouldn't post at 3 AM.

That really only brings things down to my second point, though- the developer just needs to finagle a sizeable amount of cash at one time, with the guarantee of getting it back afterwards even if he doesn't manage to rope in enough marks to double it.

you're right, entirely right. But I think this is what people want: Developers can control and fork the code at will because they have a majority (51% >) of the coins.

Coins are used for some kind of service that it provides (transaction fees for decentralized exchange, computation, storage). Developers fund development from the initial selling of the coin and have incentive to increase the price.

I wonder if the two tiered model (ETH/Gas, Factom/Entry Credit, there are lots) is better or worse

So much money grabbing happening right now, and often they don't have any kind of service or working software yet. Definitely a lot of specutlation, are these all tulips?

> But I think this is what people want: Developers can control and fork the code at will.

The only reason the blockchain was invented is to avoid having a centralized authority.

I don't agree. The blockchain enables a type of authoritarianism that I support, open data authoritaianism. Sort of like the linux kernel or open source software, but for the data. If I don't agree with the leader, I'm free to fork the ledger and build support for a new leader. Just imagine if we had this tyoe of power over facebooks "ledger", how much easier would be able to hold a check on the centralization of power as a user. I don't mind authoritarianism as long as I have the real control to take the data and elect new leadership at any time.

I agree, if they don't consider the worst case scenarios then its clear they don't consider security in mind.

I really don't understand why people, especially from cryptocurrencies, are stuck up on the fact that "Game Theory" can explain everything. Human nature is fickle. Things can happen which are beyond the coded game theory assumptions, especially when everyone is aware of the rules.

One of the biggest assumptions which has always been made is on the miner's side. They can't attack the network because 51% stops it. But that has proven wrong when F2pool manipulated the Status ICO:


So what happens if there is a miner working in conjunction with the devs?

The document sees the 51% attack originating from token holders, not miners.

Buterin is in danger of becoming a parody of himself. I literally thought this was satire for several paragraphs, before glancing up at the byline.

A DAICO is a merger of two dubious ideas, and inherits the flaws of each. It has DAO problems- that it trusts a large quantity of money to public code that can't be fixed without breaking the guarantees provided by the immutability of that same code. It also has ICO problems- the combination of extremely volatile "stores of value," non-expert enthusiasm, and anonymity of the developers is ripe for fraud.

Something like this might work for a "distributed Kickstarter," once the bugs have all been worked out of a standard "form" contract and assuming that no payout is promised to token-holders. But this requires abandoning the core conceit of an ICO (that there are valuable "coins" involved) as well as a reasonably important feature of DAOs (that they can be meaningfully adapted towards specific use-cases.)

I don't think that is necessarily true, while you might be right in the sense that more complex behaivour has the chance of introducing more bugs into the code.

The tools, frameworks and contracts keep getting better and more mature, so hopefully we won't see occurrences like the ones of TheDAO.

> But this requires abandoning the core conceit of an ICO (that there are valuable "coins" involved) as well as a reasonably important feature of DAOs (that they can be meaningfully adapted towards specific use-cases.)

I honestly don't agree with this, there is no reason for coins not to retain its value.

Vitalik initial concept is promising but I encourage everyone to read the rest of the thread there are some very good alternatives and additional ideas.

Overall, if the main goal is to improve the ICO mechanism to reduce fraud and increase accountability from the development team, that in my book is worthy of consideration and discussion.

Why would anyone give X money to a new venture if there’s no mechanism by which they get Y*X money back?

Wouldn’t it make more sense to dispurse new funds in tranches only after the first tranch has been repaid with interest?

Why would I as a founder want the tap rate be controlled via consensus? It could hinder certain purchases if they are time sensitive.

Why would you as a founder want any kind of accountability? Maybe you can raise more funding that way. In the current environment that's not really the case, but things have to change eventually.

Purely for accountability reasons. It has it advantages on establishing more trust on the investors

> Any vote is subject to 51% attacks...

Isn't the entire point of voting in the first place to determine what >= 51% of the voting population desires?

Calling a system working exactly as intended an attack seems a bit...odd.

That fat comment with the fetch a grownup button is mine ;)

Really enjoyed your contribution to that thread.

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