[2017-08-08] Dear Bitcoin: I’m Sorry, Fees Will Rise

[Emoclaw]

Bad news travels slowly, in the cryptocurrency world. Tribalism causes unpleasant facts to be treated as an attack. But if nobody tells the truth, you can’t blame anyone for being confused.

Decentralized payment systems don’t scale up; to really know if a payment to you is valid you need to know every payment which fed into it was valid. To know if the system is working you need to know nobody is faking money out of thin air, and that nobody is simply stealing money from others. To get this level of confidence, you need your own computer to check everything, and so it needs to see everything, and that’s never going to scale.

It’s a superb and unprecedented level of trustlessness. But it doesn’t scale.

Bitcoin can’t scale to meet demand, unless demand vanishes and it all fails. If bitcoin is used by 100,000 people today, and we want everyone to use it, we need bitcoin capacity to grow 75,000 times larger. Yet we’re already struggling with 140GB of storage caused by bitcoin’s first 8 years when it was mainly not being used, and rampant mining centralization which makes everyone nervous about making things more difficult for small, independent miners.

• Bitcoin fees are only going to rise from now on. Plan on it.
• The largest cause of non-spam transactions seems to be companies who designed their infrastructure assuming transactions are cheap. The big VC-backed ones know making their operations more efficient won’t increase their valuations, but expanding their business will. Their pain threshold is clearly higher than the recent fees.
• Even with segregated witness, demand can easily outpace the increase. There might be a short 50% drop in fees for early segregated witness adopters, but it won’t last.
• Even a double-once hardfork, which doesn’t require anyone to actually optimize anything, won’t give relief for long.
• Other once-offs like Schnorr similarly only offer 30% increase, easily swallowed.
• The only attempt to attach growth to a metric was Pieter Wuille’s 17.7% per year increase. That takes 70 years to reach everyone-scale, so no relief there in my lifetime.

The scaling problem was the very first criticism raised when Satoshi posted his proposal to the cypherpunks mailing list. That was a champagne problem then, but a real one now.

This is the reason I’ve been working as hard as I can on solutions, but denying the problem doesn’t help.

https://medium.com/@rusty_lightning/dear-bitcoin-im-sorry-fees-will-rise-b002b1449054

[buwaytress]

Thank you for your work on LN - others I'm sure are also working on other solutions, also concerned with fees (whether earnest concern or a professed one, such as Bitcoin Cash).

Yes, scaling was an early issue picked up on and continues to persist. But scaling was always going to be a constant that technology would always have to play catch up to. Look at the Internet, constantly grappling with its biggest issue: scaling. It was also a very early issue for implementation. It went from storing hosts on a text file to DNS. Then in the mid-90s we tried to slow down routing table growth on the Internet, and then next we were running out of IPv4 addresses so we went IPv6. Still less than half of the planet is on the Internet and routing scaling problem isn't fixed... we're patching it with bigger, faster routers but there is constant scaling tech that keeps catching up to the needs.

The same is happening for Bitcoin, with the added complexity of consensus... but the same pattern for scaling was predicted as early as 2012/3(?) by Antonopoulos.

I wonder if BTC value rises or will rise partly to keep BTC fees as stable as possible. More expensive transactions are for certain. But Bitcoin may yet represent more cost-effective txs for many years to come.

[Netnox]

This is correct. As the number of Bitcoin users increase, the number of transactions will also increase. Even if the block size is increased to 2MB, it will prove inadequate in the near future. A permanent solution is needed to resolve this issue.

[1Referee]

From the very early days, when I could still get first block confirmations without a single satoshi in fees, it was known (at least to those who understand how Bitcoin is meant to stay rewarding for miners) that fees would undoubtly increase throughout the years due to the lower subsidy rewards as result of the block halvings. Also was clear that on top of the fact that lower block rewards would eventually result in the fees to increase, we also had to keep the constantly growing demand for block space in mind, spam attakcs, etc - these all add oil to the already boiling fire. In that regard, I can only applaud initiatives such as LN to offer people a far better, faster, and most importantly, a cheaper currency experience for those who need it. I am perfectly fine with how Bitcoin is operating right now, but those who don't mind paying a higher fee, are an extreme minority here.

[soothaa]

I still do not understand this blocksize war. I have seen the posts of our Saint himself and he even said that the 1MB blocksize was a temporary measure and that 2/4/8MB would be used in the future. That was 7 years ago. Why the heck is there all this trouble simply moving to a 2MB blocksize? Is it all really simply due to the blockchain increase? Who cares?! Full nodes can run 8TB+ drives for a trivial amount of USD these days and checkpointing/snapshotting the blockchain is a completely viable thing..

If I am wrong please point me to reading that directly disproves these things..

[Carlton Banks]

Yep.

Lightning layer transactions will be the real "Bitcoin cash", as they will be settled instantly (unlike the blockchain running beneath the Lightning network, which will probably always take minutes for confirms to act as settlement).

And Rusty's main point is the real one: Lightning is a scaling solution that actually scales the Bitcoin network, lol.

[hatshepsut93]

I still do not understand this blocksize war. I have seen the posts of our Saint himself and he even said that the 1MB blocksize was a temporary measure and that 2/4/8MB would be used in the future. That was 7 years ago. Why the heck is there all this trouble simply moving to a 2MB blocksize? Is it all really simply due to the blockchain increase? Who cares?! Full nodes can run 8TB+ drives for a trivial amount of USD these days and checkpointing/snapshotting the blockchain is a completely viable thing..

If I am wrong please point me to reading that directly disproves these things..

Even with very big blocks and pruning Bitcoin won't be able to scale on-chain to the levels of payment processors like Visa. Bitcoin Cash is still a "digital gold", because it can't process 5000 transactions per second.

Running a node takes different resources - RAM, bandwidth, storage, CPU, so different users might have different bottlenecks. Since there's is economic inequality in the world, increasing requirements for running a node will push 95% of nodes to wealthy countries. This will make Bitcoin less secure and permissive, because you'll had to invest thousands of dollars every few years to get full Bitcoin experience of trustless transactions. Poor people would have to use SPV and online wallets and be vulnerable to different attacks.

[CyberKuro]

I consider bitcoin fees are expensive if more than 150 sats per byte, and my last transaction 3 days ago, just paid 112 sats per byte, very cheap I thought.
I didn't realize how long it takes to be full confirmed but maybe around an hour, still faster than bch and I didn't mind about it.
Right now there are over 29K unconfirmed transactions and I know the fees already increase, but segwit suppose to solve this problem, and we have 2 weeks before it will be activated.

[Kanapka]

This is correct. As the number of Bitcoin users increase, the number of transactions will also increase. Even if the block size is increased to 2MB, it will prove inadequate in the near future. A permanent solution is needed to resolve this issue.

But how make a decentralized blockchain with lots of transactions that is safe and is adequate for large scale?

Looks hard, not to say impossible