Hacker News Board Hacker News, browsed like a classic message board
Live data from Hacker News

An analysis of Bitcoin's throughput bottlenecks

github.com

Post reply on HN
Print view View original

141–150 of 234 posts

Re: An analysis of Bitcoin's throughput bottlenecks

#141

Post by MereInterest »

lottin wrote: post #57
djanogo wrote: post #13

Earlier quoted context omitted.

I don't know how Bitcoin transactions work, but can a company (like Paypal) buy bunch of bitcoins (10000), and then proceed to do all transaction internally with their own ledger mechanism without depending on other networks?, wouldn't this allow anybody within paypal network to send/receive super fast?

The whole selling point of bitcoin is that it's a decentralised ledger that isn't owned by anybody. If the transactions are managed by a central entity, then that's no longer true. For example, what's to stop such an entity from creating bitcoins out of nothing and thus debasing the bitcoin currency?

Absolutely nothing is preventing it. Depending on your viewpoint, it happened back in 2014, and somehow the cryptocurrency community just sort of shrugged and got back to wildly speculating.

https://en.wikipedia.org/wiki/Mt._Gox

Re: An analysis of Bitcoin's throughput bottlenecks

#142

Post by chris_wot »

dalmo3 wrote: post #65
lottin wrote: post #53

Earlier quoted context omitted.

Sorry but bitcoin is peer-to-peer cash system. If it can't handle peer-to-peer transactions then it's an utter failure.

99% of the content on news.ycombinator.com is not about startups. An utter failure!

Nice try, but that's not the point of news.ycombinator.com

Re: An analysis of Bitcoin's throughput bottlenecks

#143

Post by CryptoPunk »

billytetrud wrote: post #117
DennisP wrote: post #89

Earlier quoted context omitted.

Ethereum shards share security, so you can't just do a 51% attack on one shard. Zkrollups on Ethereum have the same security properties as on-chain transactions, with no need for monitoring and no withdrawal delays. That can't be said for the LN.

Sharding is fraught with issues. One fundamental issue is that by the very nature of sharding, you have fewer nodes on each shard, which makes it easier to perform sybil attacks on a particular shard. Shards aren't a pancia. Also, if you have, say, 10 shards - that isn't giving you 10x throughput. Its gonna give you maybe 2x throughput because what shards are doing is allowing lower-power machines help run the networ…

All nodes validate the root hash of every shard, so no, a Sybil attack on one shard is no easier than on the entire blockchain.

It may even be harder, as there would be more nodes in total when the hardware/network resources a person needs to run a node decline from what's needed to validate the whole blockchain, to what's needed validate the fraction of the blockchain contained in one shard, or conversely, when 64X more economic activity, and thus more people, are utilizing the blockchain, as a result of the scalability boost provided by dividing the blockchain into 64 shards.

>>Also, if you have, say, 10 shards - that isn't giving you 10x throughput. Its gonna give you maybe 2x throughput because what shards are doing is allowing lower-power machines help run the network, and those lower-power machines are not as effective as the rest, so they aren't contributing as much.

They don't have to be lower power machines. If the blockchain is divided into 64 shards, then the total network throughput can increase by 64X with the power requirements of validating machines not changing.

Re: An analysis of Bitcoin's throughput bottlenecks

#144

Post by XorNot »

PeterisP wrote: post #48
bourgwaletariat wrote: post #12

Earlier quoted context omitted.

How many transactions happen in the world?

Looking solely at electronic direct payments, it would be more than a billion transactions per day, there are multiple major schemes that work on the 100m/day scale each (Visa, Mastercard, US ACH, EU SEPA, China Unionpay, etc) and a lot of smaller ones that add up. So that's in the ballpark of 10k sustained average txn/sec, more in peaks. On the other hand, there's room for a lot of growth, it's not even a single tra…

Which is the core point: the major processors aren't being limited by their technology, they're being limited by the size of the economy. There's no real meaningful limit on how they can scale that.

Just barely creeping across the line with crypto means its already inadequate.

Re: An analysis of Bitcoin's throughput bottlenecks

#145

Post by tshaddox »

MereInterest wrote: post #22
djanogo wrote: post #13

Earlier quoted context omitted.

I don't know how Bitcoin transactions work, but can a company (like Paypal) buy bunch of bitcoins (10000), and then proceed to do all transaction internally with their own ledger mechanism without depending on other networks?, wouldn't this allow anybody within paypal network to send/receive super fast?

Correct. This adds a point of centralization, which defeats the entire purpose of cryptocurrency. As far as I can tell, cryptocurrency is based on wanting (1) a shared database that (2) can be updated by anyone within specific rules and (3) doesn't require anybody to trust anybody else. In the same way that low-trust societies have a much bigger overhead as a result of the lack of trust, I do not think cryptocurrenci…

> This adds a point of centralization, which defeats the entire purpose of cryptocurrency.

I don't think you get to declare what "the entire purpose of cryptocurrency" is! People tend to get confused about what "trustlessness" actually means. It certainly doesn't mean that every single aspect of every transaction involving cryptocurrency must not require any party to place any trust in another party. If you're using cryptocurrency to buy bread you still need to trust that the baker hasn't poisoned your bread, and that need for trust must be addressed by some mechanism unrelated to cryptocurrency (like the baker's reputation, or laws against poisoning people). That doesn't illustrate some absurdity in the idea of cryptocurrency or "defeat the entire purpose of cryptocurrency."

Re: An analysis of Bitcoin's throughput bottlenecks

#146

Post by bluepnume »

Take a look at "The Blocksize War" for a fantastic rundown of the events that led to Bitcoin not increasing its throughput.

TLDR: Massively rushed changes, pushed by people with a lot of power on the network (miners and corporations), which ended up being largely antithetical to Bitcoin's decentralization thesis.

Bitcoin's value is in its stability, its decentralization, and its lack of hurried change. It is in the control of everyone, not just the powerful actors in the system. Its value is not in its ability to pivot every few years like a startup to fulfill some new purpose.

Re: An analysis of Bitcoin's throughput bottlenecks

#147

Post by CryptoPunk »

billytetrud wrote: post #113
Havoc wrote: post #82

I’m still not quite following why people are looking at bitcoin for this. Surely one of the newer generation like eth is better suited? Or better yet a layer 2 on eth

What's better about Ethereum in this regard? People are still looking to Bitcoin because its substantially larger than Ethereum in pretty much every way (development, market cap, awareness, etc), its got a way better track record than Ethereum, its far more decentralized (Vitalik is still the god of Ethereum). There's lots of reasons people still look to Bitcoin. > Surely one of the newer generation like eth is bette…

There is substantially more development on Ethereum:

https://medium.com/electric-capital/electric-capital-develop...

There is also substantially more economic activity on Ethereum, with users paying 10X more fees to utilize it over the last 7 days:

https://twitter.com/iamDCinvestor/status/1392468184482164741

Re: An analysis of Bitcoin's throughput bottlenecks

#148

Post by MereInterest »

CSSer wrote: post #62
MereInterest wrote: post #22

Earlier quoted context omitted.

Correct. This adds a point of centralization, which defeats the entire purpose of cryptocurrency. As far as I can tell, cryptocurrency is based on wanting (1) a shared database that (2) can be updated by anyone within specific rules and (3) doesn't require anybody to trust anybody else. In the same way that low-trust societies have a much bigger overhead as a result of the lack of trust, I do not think cryptocurrenci…

> without massive, massive expenditure as a result. Could you elaborate on this? The problem I foresee is that having someone you have to trust is not a bad thing because at least you're able to identify that entity's traits and act accordingly. If that entity is, for example, the world's largest military power, I might feel like I have a lot less to worry about than if they're, as another example, a publicly traded…

In general, the more people cooperate, the better the results are overall. The downside is that the more people cooperate, the more they can be taken advantage of by defectors, by bad actors who decide to harm the public good in exchange for personal benefit. I absolutely agree that being able to trust particular actors is a huge benefit.

Part of the problem is language. "Trust" isn't just trust in a particular actor's intent, but also in their capabilities to carry through on a promise. In order to trust somebody to carry out a promise, you need to indirectly trust on others not to interfere with that promise. In this case, if you don't trust the US government to handle currency, then you certainly can't trust a publicly traded company that is beholding to work within and has assets that can be seized by the US government. If you assume ill-intent on the part of the government, then there's no reason to assume that a smaller actor wouldn't be influenced by the government.

In this case, I trust the US government to manage currency well, because it is in its best interest to do so.

Re: An analysis of Bitcoin's throughput bottlenecks

#149

Post by hanniabu »

rawtxapp wrote: post #66
dylkil wrote: post #23

its a pity bitcoin devs are so opposed to changing the protocol. So much research has been done showing the feasibility of increasing bitcoins throughput. Xthinner for instance is capable of compressing bitcoin blocks by up to 99% using bloom filters [1]. Much of this research was conducted on the bitcoin fork, bitcoin cash, by people ostracized from the bitcoin community for wanting to explore these ideas. [1] https…

There are some very important changes happening in bitcoin (taproot, eltoo, lightning). Devs aren't opposed to protocol changes, they are opposed to ill-thought changes that make it less decentralized and weaker.

Since when does Lightning Network not make it less decentralized? That's news to me

Re: An analysis of Bitcoin's throughput bottlenecks

#150

Post by jasonl99 »

amackera wrote: post #52
ezekg wrote: post #5

Earlier quoted context omitted.

Are there any cryptocurrencies that are attempting to solve this issue?

Algorand looks like a promising proof of stake altcoin in my opinion. https://www.algorand.com/

I agree, and I can't believe more people haven't mentioned it here. I am not going to rehash what anyone can read at the linked site, but it seems like an absolutely great project, and seems to be flying under everyone's radar.
Post reply on HN