Bitcoin Forum

For all cryptocurrencies based off the Bitcoin protocol, the block time needs to be chosen carefully in order to guarantee that nearly every node in the distributed network maintains the same view of the blockchain.  The block time needs to be, at the very least, several times larger than the maximum end-to-end propagation delay across the distributed network.

Choosing a very short block time demonstrates a lack of understanding of this requirement.  If you make block times very short, then nodes will generate blocks faster than it would take for notifications to reach nodes on the opposite end of the distributed network.  This will cause nodes within the distributed network to have inconsistent views of the blockchain.

What does this mean?  1) Lots of orphans.  2) Very long orphan blockchains.  3) Significantly reduced security for the network.

Earlier today, one of the WorldCoin mining pools (http://wdc.dontmine.me) was on an orphan blockchain for more than 300 blocks.  That would never happen for any competently designed coin.

Altcoins with very short block times have no real future.  Once they reach a certain critical mass, they will inherently self-destruct due to the large percentage of inconsistency within the distributed network.  Imagine the blockchain accidentally forking into two separate blockchains.  With very short block times, that's a real possibility.  Or imagine having a transaction confirmed, only to realize hours later that the transaction was confirmed in an orphan blockchain and needs to be rolled back and reconfirmed.  No one would realistically support such an altcoin.

If you're going to release an altcoin, do the community a favor and at least release something that appears competently designed.

coins with very short block times demonstrate the ability to take almost 100% of the rewards with enough hashrate to orphan everyone else trying to mine

This is almost certainly intended behavior by their creators.  It allows them to premine while maintaining the illusion of "fairness".

Yes, even with smaller block rewards for early mining.

It is literally a race of seconds for the first person to throw several dozen Megahash at the network.

once you have it, you control it.  Everyone else will be orphaned.

sockpuppets are common, anyone who isn't an idiot at this will use them

nobody is going to touch a coin that someone premined a massive % of coins.  There's literally 5,6 maybe even SEVEN figures of profit on the line, the people premining need to keep the illusion that it's a coin "of the people" before they dump.

This is absolutely a valid argument.  For anyone who thinks this can't happen, it happened with GameCoin.  Someone (presumably the developer) was able to override the mined blockchain with an even longer blockchain, giving all of the rewards to that one person.

No, gamecoin just used the exact same genesis block as feathercoin, and didn't bother to add new checkpoints. Someone just copy pasted the feathercoin block chain and it began propogating across the network.

Wow, that's a new level of incompetence that I wasn't prepared for

One small issue, BTC/LTC are too slow. As long as that remains a reality, we're going to have to at least experiment with lower block times :\

The only way to realistically lower the block times is to restructure the blockchain and the protocol for the distributed network.  You can't simply copy the Bitcoin protocol and paste in arbitrarily small block times, the Bitcoin protocol wasn't designed for that.

Amén!!!

That I am aware of, but hey, all of these coins are technically "experiments", so there's always time and room to do just that. Gotta start somewhere

Sounds like a load of crap to me, lets have some maths to support the claims please.

How many hops do you think represent one end of the chain to the other?

Ever heard of the 6 degrees of separation theory? The Internet is a mesh not a daisy chain.

How do you compensate for a node that has bad lag or off-line for half an hour? Well above any useful block rate.


Let me quote from section 5 of "Bitcoin: A Peer-to-Peer Electronic Cash System" by Satoshi Nakamoto


"New transaction broadcasts do not necessarily need to reach all nodes. As long as they reach
many nodes, they will get into a block before long. Block broadcasts are also tolerant of dropped
messages. If a node does not receive a block, it will request it when it receives the next block and
realizes it missed one"

I have never seen Satoshi claim that 10min was a good block rate, his only mention of 10min in that paper was in an example of how to calculate annual disk space usage of a block chain, obviously trying to make the usage no look all that much.

I agree with the OP.

These alt-coins are dying by their own hands when their block times are too short/small.

There's no long term hope for a coin with a short target block frame.

And not because of the weaker blockchain security, but rather, these coins which are more like orphan networks. Creating 10 bagillion different forks with orphans, I don't know that anything positive will come from this.

TBH I want to join some alts but I'm sticking to LTC lately, too much pump/dump coins around; I'm trying not to get sidetracked from coins which have solitude behind them.

The fact that an entire mining pool was on an orphan blockchain for more than 300 blocks isn't enough to convince you?


I'm not really sure what you're asking here.  "Maximum end-to-end propagation time" is the maximum amount of time it would take for a message from any node to reach every other node.  Each node is an "end" in that each node is both a source and destination for messages.  For a given node, the "opposite" node is the node out of all nodes where sending a message to would take the longest.


Can you articulate why that theory is relevant here?  In that theory, each person can have an unlimited number of connections.  In Bitcoin distributed networks, each node has a limited number of connections (only eight by default).  So again, can you articulate why that theory is relevant here?


The network is connected enough so that there is multiple ways to route messages between each pair of nodes.  If a node dies on a route between two nodes, messages are redirected across other routes that don't involve the dead node.

If vendors accept a coin with a short block time(like Worldcoin), it wouldn't matter how frustrated the miners got for overwhelming the network with a huge hash rate.
Traders would still trade, holders would still spend and vendors would still convert; regardless of the miners.
The death of a coin is going to depend on the adoption and infrastructure.

One can always see this as a way to discourage too many miners from throwing hashes at the network, and if there is a major acceptance of such a coin due to marketing, promotion, new software and speed.. then it will gain value and the miners will be the only ones with a problem, that is, only if they overwhelm the network.

Very good, and the same applies to nodes that can't get an update within 15sec. They are simply ignored and the updates go via other nodes.

Internet latency is much lower than when Bitcoin was first released (2009).  Should not be an issue on modern networks most nodes can ping each other within 250 ms some can do it in 30 ms

Well, for what its worth, I was still able to solo mine 544 coins at launch with a measly 1.2 mash. Yes there were a lot of orphans, but things seemed to work themselves out.

I honestly don't think the orphan rate is that critical of a flaw when you gain INSTANT TRANSACTIONS.

Let's join the 21st century people, everyone expects payment to be instantaneous in this day and age.

Let's not confuse issues.  Orphans are due to low starting difficulty not block solve time.  Once difficulty rises, often at too fast a rate, the issue becomes the opposite - coins that were designed to be fast like CNC and FTC become slow due to the fact nobody can solve a block at the current difficulty.  Until difficulty retargets, those blocks that were supposed to take 15 seconds to solve might take 4 hours.

There hasn't been any report of anyone 'losing coins' this way, except for mining orphans, which is to expected of any crypto. Even bitcoin has orphans.

10,212 world internet latency traces were under 300 ms, considered "Nominal".  Only one is presently 300 ms or over

http://www.internetweathermap.com/map

Negative. We can have our cake and eat it too. Worldcoin has proven this.

If the network stays healthy, we are good to go. It hasn't even been a week since launch!

+1

Thousands of transfers so far, no report of lost coins during transactions.. only miner issues. And network was near 1/4 of LTC rate.
But I understand the op's concern now powpow, thx.

You know like the world series  :D ::)

Someone should just make replace the block time to nanoseconds in the bitcoin source and wham bam instant transactions. I wonder why the bitcoin developers didnt figure that one out. What a bunch of noobs. ::) ::)

Try block time of one plank time

That really isn't good enough.

As I said, the block time needs to be more than several times the maximum end-to-end propagation delay.  This is needed to prevent competing nodes from generating blocks at the same time and convincing large parts of the network that their respective block is the winning block.  If competing nodes can convince large parts of the network that their respective block is the winning block, you risk fragmentation of the blockchain.  Having the block time be at least several times the maximum end-to-end propagation delays helps prevent competing nodes from convincing large parts of the network that their respective block is the winning block.

So what is the maximum end-to-end propagation delay?  Bitcoin networks are constructed by randomly connecting to IP addresses.  This means that Bitcoin networks ignore geographic locations, so end-to-end communication can cross the globe multiple times.  The network construction also does not limit the width of the network, so end-to-end communication can require numerous hops.  Add processing delays on each node and you can easily get a maximum end-to-end propagation time that is tens of seconds.

Really?  WorldCoin has proven this?  Which other coin has had an entire mining pool mining on an orphan blockchain for more than 300 consecutive blocks?


With Bitcoins you will encounter an occasional orphan, not 300 orphans consecutively.

Prove it. And what was the current difficulty?

This is an example of why democracy doesnt work for everything. You might sit down and carefully choose the parameters for your program like the bitcoin developers did only for people to come use your source and max everything out like if designing a coin was the equivalent of using gameshark to get infinite gil.

The problems were reported by miners; no one lost coins in transactions, at least none were reported.
Also, the pool that took over the blockchain (erundook's) was at about 90% of the hash power since pools just started popping up and his was the most popular at the moment. Ask him.

Fortunately, we are down to only one issue which the OP didn't bring up (powpow did) and that is, will traders lose coins. I'd say 300 blocks was a good test for that.

Prove what exactly, that the mining pool was on an orphan blockchain for more than 300 blocks?  Go read the mining pool's news for yourself: http://wdc.dontmine.me/news

This happened today, so the difficulty was either 5-ish or 8-ish.  It doesn't really matter what the difficulty was though, the pool had a quarter of the net hash rate the entire day

Block chain fragmentation is not the issue, an overhead for sure, but it's the "double spend attack" from competing chains that we are worried about, which BTW doesn't means "51% attack". You can try to double spend with less but you may not be successful. The more blocks generated, the smaller your chance.  The probability of success with a <50% attack depends on the number of blocks and not the amount of time. This is an advantage of shorter block times.

It was made open source for the very reason that it could be improved upon, in the future, when networks and computers became better.

That, and it would be harder to shut down multiple coins.

Okay, sure, you can't lose coins that weren't given to you in the first place.  The point is that the network can't stay self-consistent with the given block time parameter.

Yeah like Firefox is open source so people could change the icon image and about:config for backspace and call it "Flamefox: the ultimate opensource browser. Way better than Firefox"

Good post.

That pool is really not that good. I was mining it for a while after launch, and I had constant disconnects. I wouldn't doubt the pool operator was doing something wrong.

I've been on Big Verns pool since then, and never had a problem.

Talk to me when this has happened on multiple, more reputable pools.

Cool story bro

So what you're saying is, fragmentation is not an issue, except that it allows for double spend attacks.  Sounds like that means fragmentation is in fact an issue.

The probability of success with a <50% attack depends on you generating significantly more blocks than everyone else.  The global block time neither helps nor impedes your ability to do so.

I have used several of Paul21's pools and when they work they are great, but they tend to go up and down like a yoyo, he often says he is getting DDoS, hence not a great example to use as a short block time argument.

I seriously doubt that re-org splits from shorter block targets, create longer times to arrive at consensus on block chain validity, irreversibility is a function of number of blocks, not time.

At any rate the coins are out there, WDC/DGC, we shall soon know how they perform, no need to speculate.

Great retort. I assume you must be some type of middle school debate champ.

So tell me genius, why was it made open source then? Moral support alone?

GeistGeld has been humming along on 15 second blocks for what, years now?

So far the biggest problem was been the huge amount of RAM needed, more even than I0Coin.

I guess in a couple of years we will find out whether the RAM usage is due to the sheer number of blocks.

Or, we could update GeistGeld's code in case maybe it has a memory leak, as I0Coin is thought to have.

-MarkM-

If WorldCoin was meant to be an experiment of 'ultra-fast transactions', they should have tried a even lower block retarget time just for the hell of it.

15s has already been done several times, Geist Geld and SmallChange.

Anyway, it would be interesting to see something like satoshi DICE on 15s blocks  ;D

Do WDC say it's an experiment? No they don't, they expect it to be taken seriously.

What was the peak number of miners?

Yes, but like most things in life, an important balance between too much and too little exists. Shorter block times make it harder for <50% attacks to succeed, true, but by increasing the number of orphans, it decreases the effective hash rate of the network, making a >50% attack potentially much easier. Bitcoin considers the potential risks of a successful >50% attack to be very high, so Satoshi chose a long block time.

+1.

This seems like mathematical nonsense, please prove to me that  for example if there are 4 times more chain splits with 1/4 the block time, that there are less iterations to arrive at consensus within the same duration. I bet you can't, because the increase in chain splits is trivial compared to the increase in blocks.

LTC has certainly proven for a long time that a shorter than 10min block target is perfectly viable.

I don't think anyone can credibly claim that block times must be 10 minutes, LTC and several other coins run effectively at 1-2 minutes a block. This should hardly come as a surprise, the improvement in internet latency in western countries and processor speeds in the last 4 years would give us that kind of improvement.  In other-words LTC now is likely producing orphans at about the same rate BTC did when it first launched so if those parameters were correct for BTC at the time then they are fine for LTC.

The OP is merely claiming that trying to get as the very short times around 15 seconds is unwise and unsafe and this is a legitimate concern particularly because its very very common for new coins to run even faster then target and they are certainly going to exceed propagation speed at those times and thus we get the 300 orphan blocks and other similar effects (we had some bad orphaning during FRC launch and some large orphan forks too).  If difficulty was better constrained to match network hash rates then 15 seconds might be sustainable.

Now I do not believe that LTC or other faster block chains confidence increases at a 1:1 ratio, in other words 6 LTC confirmations != 6 BTC confirmations, you do need more confirmations on the faster chain but you should NET need less TIME.  And that's what we really care about when doing commerce the time to get verification.  It's just not a linear relationship as may advocates of fast block-chains claim, their is some kind of logarithmic scaling that's yet to be really measured well, I suspect that if we did measure it we could arrive at a scientifically optimum block-time.

Of course a block time shorter than 10 minutes is perfectly viable! I never said that it wasn't. My point was that there exists a point where the orphan rate caused by a shorter block time ends up outweighing the benefits of a shorter block time. This starts to matter earlier than you may think because an orphan rate of just a few percent would be enough to encourage miners to centralize. However, ignoring that, here's an extrapolation of the reduction in effective hash rate as a network the size of Bitcoin reduces block time (the Bitcoin network currently sees about a 1% orphan rate):

Blocktime in seconds	Reduction in effective hashrate
600	1%
300	2%
150	4%
75	8%
37.5	16%
18.75	32%
9.375	64%

As you can see, two and a half minutes per block isn't actually all that bad in the reduction of effective hashrate, but I wouldn't want to go much lower than that.

Maged:  Do you have a source for your numbers their?  Or are you just postulating that each halving of block time doubles the orphan rate?  It seems a bit too smooth to be realistic to me, miners presumably have a variety of connection speeds which should add some noise.  Do we have data on the orphan rates in LTC or Geist Geld and SmallChange which were mentioned earlier?

How does your table prove this? Where is the formula? What's this effective hash rate nonsense. We have block and timestamps on those block we can check to see if we are getting them at roughly one ever 15 sec.

If a 15sec block rate didn't work because of orphans my WDC mining stats would tell me this and they don't, in fact WDC is working really well, and is doing transactions faster than any other coin I have tried.

i think that less tahn 2 mins is a bad proposition.

it is funny that all these 'pro' came out with magic numbers that 'disproof' viability of short block times.

Yes. The propagation time does not change as the block times get smaller, and the math for those values were easier. To think of it another way, at a 6 second block time, the effective hash rate is reduced by 100% - meaning that the single fastest miner will always get the next block (if you assume that blocks are produced at a constant rate). As for the 1% number, that was what someone saw as Bitcoin's orphan rate a few months ago. I'm going with that number because it is good enough for this basic analysis.

Obviously, this is based on averages. An individual block could have a large space between blocks, allowing more propagation before the next block is found. Of course, the inverse is also true.

Doesn't matter. The Bitcoin network should dictate the absolute minimum viable block time, since it is currently the largest network.

Only thing is less than 2 minutes is what we need for these things to be viable :\ It's a battle between security and usefulness

I agree what happened with the large pool was an example of the disaster that would occur should these new coin of the months ever reach mass adoption.

We ALREADY have the alt coin we need.

LTC.  Nothing else introduced really touched it.

Now we have Feathercoin 2.0 coming - don't get me wrong I will hop on the train and rip off BTC/LTC owners again who want to buy a bunch of FTC 2.0 just like anyone else.

This is going to continue until the rubes quit.

LTC is already the coin that is fast enough while staying 100% secure and has the history and most distributed ownership by percentage of coins outstanding.