An ‘AMA’ with the Tsinghua International Blockchain Association (part two)

We were recently invited to take part in an ‘Ask Me Anything’ with the blockchain folk at Tsinghua University — during which time we took questions on everything from consensus mechanisms to the future of bitcoin. Here’s the full transcript (part two of three).

“Given your concerns about the unsustainability of permanent ledgers, what are your predictions for the future of bitcoin? Is its collapse inevitable or can you envisage a way that its scalability problems are overcome?” — Peter Y

David Lancashire: BSV is already saying miners won’t save blockchain data. BCH is putting in “UTXO commitments” (so no-one needs to sync older blocks at all). ETH is already so big it takes 1 full day to sync about 8 days worth of blocks. And BTC is holding the fort, but won’t have the fee volume to pay for security at scale. So — yeah — I’m pessimistic.

Bitcoin could implement the transient chain and that would help. But there are other (worse) economic problems. So even if they did — would it matter? For me to be optimistic, I’d need a sense that the devs didn’t have their heads in the sand on this question. But everyone is in denial about the problem. In the long-run it won’t matter if bitcoin doesn’t get its act together. It really only matters for people who are financially invested, and this is at least a 5-year issue, so it isn’t as if bitcoin is a bad investment or a critical issue for most people

“Can you elaborate on what you mean by ‘at scale’?” — Peter Y

David Lancashire:by “at scale” I mean at the limit of what is economically and technically feasible given the amount of money available in fees. The nice thing about paying for bandwidth instead of hashing is that you approach technical limits properly. 100 GBPS is reasonably available to large ISP-level actors today. We’re expecting this to increase by an order of magnitude in the next decade. Latency and block-propagation issues aside, that puts the upper limit on a blockchain somewhere around 100 TB today and maybe a petabyte in a decade. Storage issues hit us faster than bandwidth once we incentivize people paying for networking equipment.

“Context for above question: Couldn’t one argue that bitcoin doesn’t need to ‘scale’ if it functions as a secure settlement network (like the real time gross settlement system used by central banks) rather than an everyday payments system (like Visa and MasterCard)? The former requires a much lower number of transactions making scalability less of an issue.” — Peter Y

David Lancashire:you’re assuming that a small scale network can be secure… why? Miners won’t burn more in energy than they’re collecting in fees. If the average block contains 50k USD in fees, that means you need to rent something like 25k in hashpower to collect all of the fees.

You want to run an international settlements system on a network secured by 25k every 10 minutes?

Maybe to chime in something extra — this is the single biggest issue we run into with people who are used to POW and especially hardcore BTC maximalists. They assume that blockchains are secured by “hashing” or “cryptography” instead of money. So they don’t think about the economics of the network as the block reward is scheduled to die.

It is the biggest question for BTC design — will fees rise to whatever arbitrary levels are needed to keep the blockchain secure. 

Thanks for the follow-up question. hopefully my answer was coherent.

“You say: ‘You’re assuming that a small scale network can be secure… why?’ Because at the current scale at which it operates the bitcoin network is secure, never been subject to a successful 51% attack. The argument is that the current scale of the bitcoin network is sufficient to run a secure real time gross settlement system, with low value transactions taking place on centralised platforms (e.g. a WeChat Pay solution denominated in bitcoin) or second layer solutions like lightning.

“The point about what happens when the block reward runs out is an important one. Correct that transaction fees would have to rise in order to account for the energy expended to mine. However there is real value in secure settlement and this is a service people would be willing to pay for (as they have done historically with gold settlement)” — Peter Y

David Lancashire: the amount of mining on BTC is very obviously propped up by the block reward. That is going away. Why is your model for how BTC will work in 2029 based on how it has worked when it is printing free money for miners? How is this reasonable?

And tons of POW chains have had 51% attacks. If you think Bitcoin is going to be so secure without the block reward, why aren’t you using those networks? What is the difference between those networks and bitcoin? It is basically speculation, which means your entire security model is based on ponzi economics. Not a good idea.

Payment channels like LN will run on the blockchains that are large, scalable and secure. At a minimum, a small-scale Bitcoin needs to eliminate the 51% attack so that it is protected by 100% of fee volume. That can help. If you know a single BTC dev that is working on that, who?

“With incentives, stakeholders often secure decentralized networks, how do you make sure that groups that start with an advantage (by accumulating large stakes at low costs in the beginning or have access to cheap computing power) don’t use it to become dominant players in the network and get more and more rewards and squeeze out small players? How to make sure the network stays decentralized?” — Tristan

Richard Parris: Thanks Tristan. The solution is to not have a monolithic security mechanims like PoW and PoS. Saito does this by considering Transaction fees themselves the work that is needed to create blocks. This has the important effect that work that does not go in to a block, can go into the next block. This turns 51% attacks into 100% attacks. You also need to match all the outstanding work as it accumulates. We are actually doing some exciting work at the moment that ensures that attackers have to bleed money to any honest nodes in the network. The other important facet of this is to maintain openness. Which is where dPoS really falls down. Governance layers, inevitably increase cartelisation, as voting blocks naturally form.

“What are the major merging aspects of quantum computing and blockchain as of today?” — 太阳

Richard Parris: The big risk around quantum computing is that all of our previous cryptography becomes transparent. It does not break or cryptographic techniques. The disparity in the time it takes to create key pair, and to crack it remains with quantum computing.  So, crypto we are all going to have to start again if any turly massive increase in compute power is suddenly available. But, we know how to start again.

“If you already have Dapps running on your network: (1) What kind of Dapps do you have? (2) How different is the traditional coding vs coding for Dapps (3) What would be a Dapp you’d love to have on Saito but never had the chance or time to develop?” — Cris

David Lancashire: Hi Cris. #1. //apps.saito.network — demos all here. Most of our users are spending time in the arcade (//apps.saito.network/arcade). We have a lot of Twilight Struggle players.

#2 — the modules are pretty easy to write. The biggest challenge is getting your mind around the fact that they’re all peer-to-peer (not client-server). So applications run IN THE BROWSER instead of on a server, and all updates happen when you receive messages.

#3 – I want actual decentralized WeChat that uses group Diffie-Hellman key exchanges to completely avoid the need for centralized server logics, while not having a “smart contract” that can be censored. We haven’t coded this ourselves because we’re trying to focus on applications that people will really use that don’t have heavy chicken-and-egg problems, and in part because people don’t take things seriously until they have transaction volume. But I think decentralized, encrypted chat is probably the most important social tool that will come mainstream in the next five years.

We will happily support people building apps on Saito — we’ve got some people in Venezuela who are already building stuff. We have a list of existing modules here and they are pretty easy to code and can show how it is done: //github.com/SaitoTech/saito/tree/master/mods 

The weirdest app is probably a hospital login / appointment system for a South American country that has corruption issues and wants blockchain to provide transparency into how normal people get medical care and prevent hospitals from cheating the national payment system.

Read part three here.

Or go back to part one here.

Leave a Reply

Your email address will not be published. Required fields are marked *