Hacker News new | past | comments | ask | show | jobs | submit login

Full disclosure: I work on the cryptocurrency in this article, Algorand.

There are a lot of questions and speculation here about this paper and Algorand. I would be happy to try an answer them to your satisfaction. Some context may be helpful first, though. This paper is an innovation about one aspect of our technology. Algorand has a very fast consensus mechanism and can add blocks as quickly as the network can deliver them. We become a victim of our success. The blockchain will grow very rapidly. A terabyte a month is possible. The storage issue associated with our performance can quickly become an issue. The Vault paper is focused on solving this and other storage scaling problems.

The Algorand pure proof-of-stake blockchain and associated cryptocurrency has many novel innovations aside from Vault. It possesses security and scalability properties beyond what any other blockchain technology allows while still being completely decentralized. Our website, algorand.com, and whitepaper are great places to start to learn more.

If you learn best from videos then I suggest you watch Turing award winner and cryptographic pioneer, Silvio Micali, talk about Algorand: https://youtu.be/NykZ-ZSKkxM. He is a captivating speaker and the founder of Algorand.

OK I'll bite. Nothing about the article or YT video is innovative, some of it is laughable.

Algorand's consensus per the YT video you linked is step 1. 1 user is randomly chosen to propagate a new block (can user make up a fake block? let's assume not) step 2. 1000 users randomly selected vote on that block and if they agree, it's DONE. (what if 1000 users are in the same country/company, etc. how do you prevent collusion on fake blocks)

This is hilarious. So the only thing you rely on for consensus is 1001 random 'users' input weighted by stake? What happens when your network is bombed with 1000tps and you need to contact 1001 staked users for every block? There are also vulnerabilities such as the recent "fake stake" bug that affect pure POS coins which I won't go into.

Now about the article referencing Vault. I believe NANO came up with this bootstrapping feature:

> Each user account only ever stores the balances of the accounts in its assigned shard

Yes they are replicating NANO's block lattice structure using "shards" with an insecure way of trusting any future chain that includes your old tx's but gives no guarantee on the state of any other account. As I understand it, I can be fed a fake chain while bootstrapping and accept phony funds as long as my balance shows up, right? ...Assuming this feature worked without any security holes, what's to stop other coins from implementing it? You don't need an entire cryptocurrency for it, it's just a feature and if it worked everyone would be using it. The sharding problem has not yet been solved and there are coins like Ethereum that are trying really hard to make it happen.

It's really sad to see this low-quality content come out of MIT. This looks like someone trying desperately to get a piece of the Crypto pie using MIT's reputation as a get rich quick scheme. Just for kicks I looked through reddit/r/cc and found only 5 dead posts mentioning Algorand with no comments on any of them. Their website makes all these claims about "pioneering a sortition algorithm" that looks "totally legit" if you ask me.

Can you provide more detail because from the article this doesn't seem so impressive.

> Vault reduced the bandwidth for joining its network by 99 percent compared to Bitcoin and 90 percent compared to Ethereum, which is considered one of today’s most efficient cryptocurrencies

1) Since when are these considered efficient? I don't think anybody in the know would say this. They're the most popular, but no means the most efficient.

Bitcoin is 250GB, so 90% of that is still 25GB to join the network, which is still ahhh enormous amount. And what's the baseline for comparison here? Were all of bitcoin's transactions replayed on an Algorand test network for this comparison? Or is this a metric from some test usage? If the latter then that's a huge issue since it grows in size.

2) On top of that you're saying it could accrue a terabyte a month in data. What type of usage is this under? Is that on current bitcoin transaction levels? 2017 transaction levels? A steady state tx/s? Is a backup of this data needed or is it throwaway and summarized in the latest blocks? If it's still needed then that's a decentralization issue because not many people will be maintaining full nodes.

3) What type of specs are affected by these changes? Can you still perform atomic swaps? That's a pretty standard requirement nowadays and would hinder the Blocknet and exchange interiperability.

Are there reasons that e.g. Bitcoin and Ethereum and Stellar could not implement some of these more performant approaches that Algorand [1] and Vault [2] have developed, published, and implemented? Which would require a hard fork?

[1] https://www.algorand.com/

[2] https://dspace.mit.edu/handle/1721.1/117821

My understanding is that PoS approaches follow normal byzantine agreement theory which states that adversaries cannot control more than 1/3rd of the accounts (or money in the case of algorand). You can also delay new blocks more easily.

Ethereum is scared or that so they are implementing some hybrid form.

Bitcoin is doomed from my perspective, because of the focus on proof of work and the confirmation times. When you realize that algorand is super fast, there is no "confirmation time", and there is no waste in energy to mine, then it is hard to back up any cryptocurrency focusing on proof of work.

And what of decentralized premined chains (with no PoW, no PoS, and far less energy use) that release coins with escrow smart contracts over time such as Ripple and Stellar (and close a new ledger every few seconds)?

> Algorand has a very fast consensus mechanism and can add blocks as quickly as the network can deliver them. We become a victim of our success. The blockchain will grow very rapidly. A terabyte a month is possible. The storage issue associated with our performance can quickly become an issue. The Vault paper is focused on solving this and other storage scaling problems.

What prevents a person from using a chain like IPFS?

Ethereum Casper PoS has been under review for quite some time.

Why isn't all Bitcoin on Lightning Network?

Bitcoin could make bootstrapping faster by choosing a considered-good blockhash and balances, but AFAIU, re-verifying transactions like Bitcoin and derivatives do prevents hash collision attacks that are currently considered infeasible for SHA-256 (especially given a low block size).

There was an analysis somewhere where they calculated the cloud server instance costs of mounting a ~51% attack (which applies to PoW chains) for various blockchains.

Bitcoin is not profitable to mine in places without heavily subsidized dirty/clean energy anymore: energy and Bitcoin commodity costs and prices have intersected. They'll need any of: inexpensive clean energy, more efficient chips, higher speculative value.

Energy arbitrage (grid-scale energy storage) may be more profitable now. We need energy storage in order to reach 100% renewable energy (regardless of floundering policy support).

Ripple is not decentralized. I don't know enough about Stellar to answer.

Bitcoin is software and can easily implement these features but the community is divided and can't reach consensus on anything. Lightning Network as layer two solution is pretty good from what I know.

Ethereum improvements are coming along very slowly and that's good. They're the only blockchain with active engagement by thousands of multiple parties.

Aragaon and Vault's papers might sound good, but who knows how they'll turn out in production.

People argue this all day. There's a lot of FUD.

Ripple only runs ~7% of validator nodes; which is far less centralized control than major Bitcoin mining pools and businesses (who do the deciding in regards to the many Bitcoin hard forks); that's one form of decentralization.

Ripple clients can use their own UNL or use the Ripple-approved UNL.

Ripple is traded on a number of exchanges (though fewer than Bitcoin for certain); that's another form of decentralization.

As an open standard, ILP will further reduce vendor lock in (and increase interoperability between) networks that choose to implement it.

There are forks of Ripple (e.g. Stellar) just like there are forks of Bitcoin and Ethereum.

From https://ripple.com/insights/the-inherently-decentralized-nat... :

> In contrast, the XRP Ledger requires 80 percent of validators on the entire network, over a two-week period, to continuously support a change before it is applied. Of the approximately 150 validators today, Ripple runs only 10. Unlike Bitcoin and Ethereum — where one miner could have 51 percent of the hashing power — each Ripple validator only has one vote in support of an exchange or ordering a transaction.

How does your definition of 'decentralized' differ?

How does the speed and scalability of Algorand compare to something like nano?

How do you plan on addressing the nothing at stake problem?

What do you think of the lack of privacy of algorand, compared to something like Zcash or Monero?

Zcash, monero, dash, mimblewimble and quisquis have big challenges. It is not trivial to solve the encrypted coins problem because:

- it is very hard to audit the chain for bugs. If someone finds a bug to create coins through thin air you probably won't notice it.

- regulations by states is made hard. If you are required to pay taxes, and you live in a society, then these things do matter.

If you want cryptocurrencies to work, we need cryptocoins that are not encrypting transactions.

A terabyte a month! Wow. At this point, it seems like reducing this for most clients would help. Have you looked into recursive zkp like what coda is doing?

Q: How erosive is it to the image of cryptocurrencies that there's so many coins? It seems that every week there's a new coin that does XYZ better. Often, it has no infrastructure (hardware or software) other than a limited number of exchanges that are used as speculation.

Is that really a good question to ask someone who is pushing the state of the art in this domain?

Yes - isn't that obvious? You need infrastructure to use a 'state of the art' currency as a medium of exchange.

Without the marginal energy cost of mining blocks from proof of work, what do you think would drive the value of such a currency?

This question seems to assume that the cost of electricity and mining sets cryptocurrency prices, but I'm not sure why that would be the case. I think it's the other way around. Demand for Bitcoin doesn't increase when electricity gets cheaper / mining becomes more profitable. Mining becomes more profitable when demand goes up. If electricity became more expensive or mining otherwise became less profitable, demand for bitcoin does not go down; most traders and users of bitcoin probably don't even notice. The only direct effect is that some miners stop mining (and the network becomes a little more vulnerable to 51%-attacks).

Another way to think about it: if Bitcoin somehow worked without mining, that's no reason for demand or usage of Bitcoin to go down. The value of bitcoins comes from network effects and scarcity.


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