TPS Counter: help debunk the "7 tps" myth \ stacker news ~bitcoin

TPS Counter: help debunk the "7 tps" myth

2280 sats \ 43 comments \ @supertestnet 22 May bitcoin

Some people like to spread the myth that bitcoin's blockchain can only process 7 transactions per second. example fud

I made a little webpage to help debunk this myth whenever you see someone say this nonsense:

https://supertestnet.github.io/tps-counter/

For example, today the block with the highest TPS count (so far) is this one which had 7582 transactions in it, which means bitcoin processed about 13 tps (7582 divided by 10 minutes divided by 60 seconds = 7582 / 10 / 60 = 12.6 tps).

view all related items

388 sats \ 6 replies \ @elvismercury 22 May

I applaud your effort to put things like this to the test, but I don't consider this FUD. Whether it's 7 or 13, it's broadly the same -- L1 is extremely limited, we should know that, and figure out alternatives.

122 sats \ 5 replies \ @supertestnet OP 22 May

I don't consider this FUD. Whether it's 7 or 13, it's broadly the same

It is a bit like saying "This swimming pool cannot hold a great white shark because it is only 7 feet deep" even though the deep end of the pool is 13 feet deep. The original point is true (a great white won't fit) but the supporting evidence is false.

If the critics merely said "bitcoin's L1 cannot support 7 billion people" I would agree with them. But when they back up that fundamental truth with the false claim that bitcoin can "only" do 7 tps I think their error is worth highlighting.

0 sats \ 4 replies \ @didiplaywell 22 May

Your argument is false too. It's like saying "i didn't stole money! you said I stole 100, but in reality I stole 99, so as your argument is false, I'm innocent!" The point raised by elvis is clear and the reason L2 exists!

100 sats \ 3 replies \ @supertestnet OP 22 May

Your argument is...like saying "i didn't stole money! you said I stole 100, but in reality I stole 99, so as your argument is false, I'm innocent!"

Imagine I argued that bitcoin is scalable on L1 because its max tps number is closer to 12 than 7. If that was my argument, your analogy would hold. But that's not my argument. I agree that bitcoin is not scalable on L1, I am just pointing out that the FUDers massively understate its capabilities there.

In service of that point, I am doing something similar to the part that says "you said I stole 100, but in reality I stole 99," except for these very significant differences:

(1) my numbers have a more dramatic disparity. Your analogy has a 1% difference where I'm highlighting a more than 40% disparity
(2) If a prosecutor accuses you of stealing 100 gold bars when the facts clearly show that fewer than 60 gold bars are missing, that is something you ought to mention to the judge, because overstating the crime might result in overdoing your penalty (or in bitcoin's case, understating its L1 scalability might result in devs not utilizing all of the space they can actually work with)
(3) Even if bitcoin's actual max TPS number is low, that does not justify pretending it's over 40% lower than it is in reality

0 sats \ 2 replies \ @didiplaywell 22 May

My analogy is correct in that you said "If they back up their argument with the false claim that bitcoin can only do 7 tps, when in reality it can do 13, they are wrong". I understand the relative proportions but it clearly makes no difference in practical terms and that's the reason L2s exist. My point is that the entirety of the argument is not worth of discussion regarding the FUD but the FUD itself makes no sense to start, because it's criticizing something that's designed to work the way it does. The FUD criticises big cargo ships for not going at Mach 20 when it clearly haves a distinct function.

0 sats \ 1 reply \ @supertestnet OP 22 May

it clearly makes no difference in practical terms

I think it makes a difference. When designing systems to work within bitcoin's limits, it is important to know what those limits are. There might be a system design that works as long as most blocks can hold 7000 transactions, but breaks if blocks are limited to 4000 transactions. Therefore, a person with this false 7 tps idea in their head might discard a very useful design simply because of false statistics. That would be bad.

0 sats \ 0 replies \ @didiplaywell 22 May

In no way I expect a professional with the technical understanding on how this systems work internally to make a final evaluation at face value alone based on blog made by an illiterate and not by checking the actual doc. But I share your fear in that sometimes companies leave this delicate selections in the hands of illiterate MBAs... like engine selections for new Boeing models...

127 sats \ 0 replies \ @03365d6a53 22 May

7 or 13 is basically the same

100 sats \ 0 replies \ @nerd2ninja 22 May

We have a feature to show the transactions per second over a timeframe on a chart in the bitcointech.help discord.

If you do a 12 month chart you see 20 TPS in some areas.

159 sats \ 1 reply \ @kruw 22 May

Your tool just calculated an average of 6.65 tps over the past 24 hours, so the claim doesn't seem like that much of a myth.

0 sats \ 0 replies \ @nerd2ninja 22 May

Averages are not max capability.

Also, we obviously have a lot of block space getting used by things that aren't transactions right now so...

21 sats \ 11 replies \ @02bcd3eeb0 22 May

According to Clark Moody's dashboard, the rate over the past 30 days has been:

[ 6.3 tps ]

So 7 does not seem like a myth at all, it seems totally accurate.

0 sats \ 10 replies \ @supertestnet OP 22 May

An average is not a maximum. When someone says bitcoin "only" does 7 tps they are wrong because it does a lot more than that on a daily basis. It is like saying a pool is only 7 feet deep when that's only an average and the deep end is 13. The word "only" (and its synonyms) is what makes the difference.

31 sats \ 9 replies \ @02bcd3eeb0 22 May

Lol, what a pointless discussion. So when we happen to get 3 full blocks within 2 minutes is that 100 tps?

When bitcoin blocks are found is probabilistic, it only makes sense to talk about averages. There is no such thing as bitcoins instantaneous tx rate.

10 sats \ 8 replies \ @supertestnet OP 23 May

I count each block as if it took 10 minutes

0 sats \ 7 replies \ @02bcd3eeb0 23 May

So then that implies what you are counting is maximum txs per block not per unit time (which would require taking an average)

0 sats \ 6 replies \ @supertestnet OP 23 May

I average the time because that is not something miners or users can control

How much block space users purchase is controllable, so devs can use that information to design for maximum efficiency

0 sats \ 5 replies \ @02bcd3eeb0 23 May

Blocks are full now and have been for months. i.e. users have been purchasing all the space. Apart from the few empty blocks that get mined they are as close to 4M weight units as possible. I'm not seeing why it doesn't make sense to say Bitcoin maxes out at 7tps, when that's what we observe empirically.

0 sats \ 4 replies \ @supertestnet OP 23 May

Blocks are full now and have been for months

But each user uses block space inefficiently, consuming many inputs and creating many outputs without batching their transactions together with other users. Fixing this through software can increase our throughput closer to bitcoin's maximum, which is far above 7 tps, as demonstrated by the fact that so many blocks do so much better than that.

On a related note:

The average tps in the past 24 hours was 7.65 tps, and 82 blocks in the past 24 hours had more than 7 tps. One even reached 11.95 tps.

view replies

42 sats \ 0 replies \ @shafemtol 22 May

The FUD is that TPS is what matters. A transaction could be anything from a single payment to a massive coinjoin. Usable capacity should be measured in terms of transfers rather than transactions.

If we define a coin transfer as spending one coin and creating one new coin, then the coin transfer capacity is roughly:

1000000 / 600 / (input_vsize + output_vsize) coin transfers per second

Assuming all P2WPKH, that's:

1000000 / 600 / (68 + 31) = 16.8 ctps

And assuming all Taproot:

1000000 / 600 / (57.5 + 43) = 16.6 ctps

That is the on-chain transfer capacity if we assume a constant UTXO set size.

In the future, full-aggregation CISA can save up to 16 vB per input, giving:

1000000 / 600 / (57.5 - 16 + 43) = 19.7 ctps

For getting new users into Bitcoin, the real limit is the rate at which new UTXOs can be made. The size of a Taproot or P2WSH output, such as that needed for a Lightning channel, is 43 vB. Thus the limit for the number of outputs per second is:

1000000 / 600 / 43 = 38.8 outputs per second

Input/output sizes from Bitcoin Optech Transaction size calculator

0 sats \ 0 replies \ @notgeld 23 May

People are always ignoring multi inputs and multi outputs transaction nature. LN tps even harder to explain but it may be estimated in millions transaction per second.

https://blog.satsbridge.com/the-most-used-smart-contract-of-all-times-2f749428adb7

0 sats \ 0 replies \ @nout 22 May

7tps is a reasonable estimate of the throughput if we consider the reality that exchanges will do batch transactions. Agreed that there's actual maximum implied by how many vBs you can fit into a block and that's closer to the 13 tps, but that's not that useful factoid. Also in the block that had 20tps the transactions are empty...

0 sats \ 0 replies \ @quark 22 May

But the average is around 7. Right? Ah yes indeed I see the average in the results in the site. You may want to cache the results

0 sats \ 0 replies \ @WeAreAllSatoshi 22 May

I didn't know about this fud until this post, though I did know that there is a large concern about txn volume on the base layer.

As others have noted, it definitely seems important to call out that the fud typically asserts that 7tps is an upper limit, not an average throughput. Definitely different measures.

Thank you for making and sharing this!

0 sats \ 5 replies \ @025738dda8 22 May

What is the myth on it? I thought that we all know that this is true. That was one of the subjects of the block size war... That's why the L2 is being built. What am I missing?

0 sats \ 4 replies \ @supertestnet OP 22 May

What is the myth on it?

The myth is that bitcoin can only do 7 tps

I thought that we all know that this is true

It's false and the blockchain proves it wrong on a daily basis

That was one of the subjects of the block size war

It was a significant talking point, but that doesn't make it true. Bitcoin occasionally exceeded 7 tps even back then, and it does so pretty much daily now that segwit increased the blocksize

0 sats \ 3 replies \ @025738dda8 22 May

So, if now there would be posts talking about 12 tps, everything becomes fine? What's the difference?

0 sats \ 2 replies \ @supertestnet OP 22 May

What's the difference?

About 40%

if now there would be posts talking about 12 tps, everything becomes fine?

If they merely started saying bitcoin can only do 12 tps, then I would have to point out that 12 tps is also not the maximum -- bitcoin also occasionally does more than 12 tps

0 sats \ 1 reply \ @025738dda8 22 May

Ok, not 12, but 13, or even 14? What is the theoretical limit now?

40 % seems to be a lot, but it is not the number that such FUDs would appreciate, I think.

0 sats \ 0 replies \ @supertestnet OP 22 May

What is the theoretical limit now?

I don't know what the theoretical maximum is but this block contains 12,239 transactions which means bitcoin briefly processed over 20 transactions per second (12239 / 60 / 10 = 20.39)

0 sats \ 8 replies \ @random_ 22 May

I think your gripe is the 'only' part. What they leave out is the 'on average' part.

Looking forward to your next project.

0 sats \ 7 replies \ @nerd2ninja 22 May

Leaving out "on average" is a big deal. This number gets used everywhere as a max, not an average. Even papers written by Bitcoiners talking about L2 proposals will reference 7 TPS as a max, not an average.

The difference is what it can do vs what people are actually paying for in a given time frame.

0 sats \ 6 replies \ @random_ 22 May

I wonder how many dev hours went into this project proving a point to noone and if they could have been better spent.

217 sats \ 5 replies \ @supertestnet OP 22 May

it took about 45 minutes

0 sats \ 1 reply \ @0fje0 22 May

I can understand some of the criticism you're getting on this "project", but I for one wouldn't want you to stop doing these.

For those that haven't seen Super in action, go watch a video of him doing his thing and you'll probably realize this was 45 min well spent.

0 sats \ 0 replies \ @random_ 22 May

I never said anyone should stop doing these kind of projects. I'm sure it was a good exercise, and now everyone has another resource that keeps track of tps. (like x)

Super is a talented badass, just like MJ.

FUD (Swackhammer in Space Jam) is the enemy, and you gotta pick your battles. Here, Super actually proves the 'myth'.

Beyond that, there's a lot of nuance. Bitcoin throughput doesn't have a limit. How many tps did ASICs mine before difficulty caught up? (because there's a limit to how much the difficulty changes)