rpietila Altcoin Observer

[fluffypony]

I do have a question about this.  (I'm very very simpleton in crypto terms - so please don't eat me)

I understand in theory the blockchain shouldn't be an issue.  Even cryptonote (I was reading about how the blockchain when it grew to BTC size would be 10X as big as BTC).  But as a consumer downloading a 20GB database is a fairly big deal with my internet connection.  (I understand I can use litewallets that sync with other servers)

But isn't the blockchain size fairly relevant to keeping decentralization in the sense that bitcoin was originally invented - to be duplicated across hundreds of thousands of consumer machines?  Even at 20GB - with my internet connection I really don't want to download the entire blockchain ...

Yes - as it stands it's definitely too large for consumer machines. The thing is, even something like mini-blockchain is only a temporary fix, as any cryptocurrency that scales up to meet massive global demand will find that home Internet connections (even in 5-10 years time) will simply not have the low-latency and speed required to maintain a full node. I remain unconvinced that there exists a better solution for "Visa/Mastercard scale" than off-chain transactions that are settled back on the main chain on-demand or periodically. A company like PayPal or Google could ostensibly operate something like this, although it's not inconceivable for Visa or Mastercard themselves to provide this.

[smooth]

I think it's a valid point but compared to something like mining centralization it's relatively minor. It's much easier to have thousands of full nodes all over the world running on VPSs and enthusiast machines than to actually have a healthy distribution of miners.

I know you didn't ask about miners but even having 1000 people running full nodes that enable people to use thin clients is still pretty good while I see mining as the major bottleneck that causes the most centralisation. Much more than lack of node distribution I think.

Having a million nodes would be good too though. A million people running full nodes solo mining on their cell phones would be pretty decentalised.

There are problems with both. Mining at least, has some incentive. There is currently zero incentive to run a node at all, and the number of nodes has been dropping even as usage has increased.

The best answer I've heard for why anyone should run a node at all is that businesses such as Coinbase or Bitpay will want to support the network since their business depends so heavily on it. This is still relying on questionable economics, but even if viable, still sounds pretty centralized to me.

The two ideas I've seen to fix this are:

1. Requiring that people run a node in order to mine (and ensuring that at least part of mining remains decentralized). That could possibly work, but I haven't seen anything approaching a solid proposal.

2. Some sort of proof-of-bandwidth where you are directly rewarded for running a node. This is even less well developed than 1.

[fluffypony]

You two really think that Gavin et al are ever incorporating any major features that are found in an alt in to their 8 billion dollar baby?

If it's not something that solves a direct threat or major problem with Bitcoin I would put my money on 'never'. Or at least not in the next 5-10 years.

Everyone always says it would be easy to for Bitcoin to do xyz, but in reality even the simplest things are bogged down in politics.

That's not to say that the Bitcoin's devs aren't doing a great job at what they set out to do: Develop a stable codebase. Competing in the market as a product isn't how I think Bitcoin is run right now. That could change of course, but its not likely to for a long time if ever.

Something interesting that's happening is that alternative Bitcoin implementations are popping up, from Conformal's btcd to Bits of Proof's enterprise Bitcoin server and so on. If there is a compelling feature and enough demand for it, and the Bitcoin developers are too narrow-minded/resolute/whatever, one of those alternative implementations will offer it. Even better - a lot of the seemingly "revolutionary" stuff being done in many altcoins can be implemented in an alternative client due to the flexibility of Bitcoin's pay-to-hash system, and existing mining pools can continue running the "regular" Bitcoin implementation and mine those transactions without even knowing about their cleverness.

[darkota]

Simply, any coins are are Bitcoin forks, will fail. Anything noteworthy on an altcoin, like mini-blockchain, Bitcoin can/may incorporate in the future, leaving said altcoin in the dust.

Seconded - 99.9% of the "features" out there can be incorporated into Bitcoin's codebase in a heartbeat.

Stealth addresses from the genesis block on (an important privacy aspect) are one of the few things they can't do, not without a roll-over to a new genesis block with all outputs being reissued to stealth addresses computed from existing privkeys (read: practically impossible).

For everything else, there's MasterCard Bitcoin.

You two really think that Gavin et al are ever incorporating any major features that are found in an alt in to their 8 billion dollar baby?

If it's not something that solves a direct threat or major problem with Bitcoin I would put my money on 'never'. Or at least not in the next 5-10 years.

Everyone always says it would be easy to for Bitcoin to do xyz, but in reality even the simplest things are bogged down in politics.

That's not to say that the Bitcoin's devs aren't doing a great job at what they set out to do: Develop a stable codebase. Competing in the market as a product isn't how I think Bitcoin is run right now. That could change of course, but its not likely to for a long time if ever.

Yes he might, there was recently a thread on the Bitcoin Discussion board about this. Bitcoin would incorporate features that are noteworthy to its cause in the future(so no anonymous features, but mini-blockchain would def be on the list)

Given Satoshi's original writings on Bitcoin you would think that anonymity would be considered noteworthy to Bitcoin's cause.

No, Bitcoin is being pushed as a very transparent, open system,. Privacy features wont be added to the bitcoin core, its all 3rd party(like darkwallet, mixing/coinjoin sites/services, stealth addresses), which sucks because as we've seen, there are A Lot of scammers in the bitcoin world and the owners of darkwallet and many mixing services for bitcoin could be/are corrupted as well.

[Este Nuno]

You two really think that Gavin et al are ever incorporating any major features that are found in an alt in to their 8 billion dollar baby?

If it's not something that solves a direct threat or major problem with Bitcoin I would put my money on 'never'. Or at least not in the next 5-10 years.

Everyone always says it would be easy to for Bitcoin to do xyz, but in reality even the simplest things are bogged down in politics.

That's not to say that the Bitcoin's devs aren't doing a great job at what they set out to do: Develop a stable codebase. Competing in the market as a product isn't how I think Bitcoin is run right now. That could change of course, but its not likely to for a long time if ever.

Something interesting that's happening is that alternative Bitcoin implementations are popping up, from Conformal's btcd to Bits of Proof's enterprise Bitcoin server and so on. If there is a compelling feature and enough demand for it, and the Bitcoin developers are too narrow-minded/resolute/whatever, one of those alternative implementations will offer it. Even better - a lot of the seemingly "revolutionary" stuff being done in many altcoins can be implemented in an alternative client due to the flexibility of Bitcoin's pay-to-hash system, and existing mining pools can continue running the "regular" Bitcoin implementation and mine those transactions without even knowing about their cleverness.

That's pretty interesting. I hadn't seen those alternative implementations before.

[itod]

You two really think that Gavin et al are ever incorporating any major features that are found in an alt in to their 8 billion dollar baby?

If it's not something that solves a direct threat or major problem with Bitcoin I would put my money on 'never'. Or at least not in the next 5-10 years.

Everyone always says it would be easy to for Bitcoin to do xyz, but in reality even the simplest things are bogged down in politics.

That's not to say that the Bitcoin's devs aren't doing a great job at what they set out to do: Develop a stable codebase. Competing in the market as a product isn't how I think Bitcoin is run right now. That could change of course, but its not likely to for a long time if ever.

Something interesting that's happening is that alternative Bitcoin implementations are popping up, from Conformal's btcd to Bits of Proof's enterprise Bitcoin server and so on. If there is a compelling feature and enough demand for it, and the Bitcoin developers are too narrow-minded/resolute/whatever, one of those alternative implementations will offer it. Even better - a lot of the seemingly "revolutionary" stuff being done in many altcoins can be implemented in an alternative client due to the flexibility of Bitcoin's pay-to-hash system, and existing mining pools can continue running the "regular" Bitcoin implementation and mine those transactions without even knowing about their cleverness.

There's nothing "narrow-minded/resolute/whatever" in being a bit conservative in Bitcoin development. It's not some unimportant altcoin whose failure will bother only a few, it's the core of cryptocoin technology. It's designed to last for a long, long time. If some feature in some altcoin stands the test a time for a year or two, then it should be considered wise to implement it in Bitcoin. IMHO this Gavin's Bloom filter improvement is mindless rush to improve the Bitcoin, it should be tested in some altcoin for some time before upgrading Bitcoin's blockchain processing. If no altcoin developer implements it first, Bitcoin developers should invent an altcoin just to test this feature.

[tacotime]

Gavin says 7 TB, certainly not petabytes, and he says his home internet connection could theoretically handle 45k transactions per second (though currently monthly bandwidth limits would prevent this -- in the future who knows).

Scroll to the bottom:

https://gist.github.com/gavinandresen/e20c3b5a1d4b97f79ac2

That's for unidirectional transmissions to you only -- in reality, you're broadcasting tx to thousands of peers and receiving them from thousands of peers, and are constantly being hit by requests both to and from you. With retransmission and other P2P ineffiencies with many peers at that network speed the bandwidth usage is probably in petabytes.

[tacotime]

thank you for the explanation .. will cryptonote bandwidth be 1/1 on storage requirement increases per transaction over bitcoin?  (Bandwidth requirements = 10X bitcoins)?

CN coins will likely require about an order of magnitude more bandwidth than Bitcoin, yeah.

[smooth]

Gavin says 7 TB, certainly not petabytes, and he says his home internet connection could theoretically handle 45k transactions per second (though currently monthly bandwidth limits would prevent this -- in the future who knows).

Scroll to the bottom:

https://gist.github.com/gavinandresen/e20c3b5a1d4b97f79ac2

That's for unidirectional transactions only -- in reality, you're broadcasting tx to thousands of peers and receiving them from thousands of peers, and are constantly being hit by requests both to and from you. With retransmission and other P2P ineffiencies with many peers at that network speed is probably petabytes.

No that's not right. On average every peer receives and transmits each transaction once. Unless you are some kind of supernode (i.e. not a home connection) you won't be getting anywhere near petabytes. On a home connection (sort of the opposite of a supernode) you will likely just receive the transaction once and not send any appreciable fraction of them to any of your peers because they will likely already have them by the time you do.

Yes there will be some inefficiency here, but not (even close to) 100x.

[tacotime]

No that's not right. On average every peer receives and transmits each transaction once. Unless you are some kind of supernode (i.e. not a home connection) you won't be getting anywhere near petabytes. On a home connection (sort of the opposite of a supernode) you will likely just receive the transaction once and not send any appreciable fraction of them to any of your peers because they will likely already have them by the time you do.

Yes there will be some inefficiency here, but not (even close to) 100x.

Right.. but that implies supernode centralization, does it not?

[blaaaaacksuit]

thank you for the explanation .. will cryptonote bandwidth be 1/1 on storage requirement increases per transaction over bitcoin?  (Bandwidth requirements = 10X bitcoins)?

CN coins will likely require about an order of magnitude more bandwidth than Bitcoin, yeah.

Is there no way to theoretically mitigate this issue or is it inherent to the tech?

[smooth]

No that's not right. On average every peer receives and transmits each transaction once. Unless you are some kind of supernode (i.e. not a home connection) you won't be getting anywhere near petabytes. On a home connection (sort of the opposite of a supernode) you will likely just receive the transaction once and not send any appreciable fraction of them to any of your peers because they will likely already have them by the time you do.

Yes there will be some inefficiency here, but not (even close to) 100x.

Right.. but that implies supernode centralization, does it not?

That's sort of inevitable, just because some connections will be faster than home connections.  (I wasn't implying a different "type" of node, just that faster connections will naturally be responsible for doing more of the distribution.)

This is not necessarly to the analysis though. If we assume every connection and node is exactly the same, then every node receives and transmits each transaction once. The total bandwidth usage doubles -- again nowhere near petabytes, even factoring in inefficiency. You need upstream matching your downstream (since you are responsible for sending to one peer on average), which limits peak transaction rates in Gavin's example to probably 4500 tps instead of 45000 tps (since home connections that run about 100mb downstream run about 10mb upstream). That's still a lot.

Also, 1 gb home connections are coming. A few people already have them, many more will in several years. Given Moore's Law this business of sending a few tiny 2kbyte transactions around is quickly becoming somewhat trivial. Gavin is on the right track.

[fluffypony]

There's nothing "narrow-minded/resolute/whatever" in being a bit conservative in Bitcoin development. It's not some unimportant altcoin whose failure will bother only a few, it's the core of cryptocoin technology. It's designed to last for a long, long time. If some feature in some altcoin stands the test a time for a year or two, then it should be considered wise to implement it in Bitcoin. IMHO this Gavin's Bloom filter improvement is mindless rush to improve the Bitcoin, it should be tested in some altcoin for some time before upgrading Bitcoin's blockchain processing. If no altcoin developer implements it first, Bitcoin developers should invent an altcoin just to test this feature.

What's wrong with testnet?

At any rate, like I said: "if there is a compelling feature and enough demand for it". I'm not talking about a "chat client" built into Bitcoin Qt. I'm talking about something fundamental like the stealth address rollover. If it is, for instance, revealed that the NSA has provably unmasked 80% of the Bitcoin addresses in use, don't you think that there will be just this sort of clamour? And in that event, if the Bitcoin developers drag their feet, Conformal and Bits of Proof will implement it and release a drop-in replacement for merchant/automation systems, et voila - the blockchain can roll over, existing miners can continue mining without upgrade/moving (unless they want to), and the world continues without needing Bitcoin Core to take point.

[smooth]

thank you for the explanation .. will cryptonote bandwidth be 1/1 on storage requirement increases per transaction over bitcoin?  (Bandwidth requirements = 10X bitcoins)?

CN coins will likely require about an order of magnitude more bandwidth than Bitcoin, yeah.

Is there no way to theoretically mitigate this issue or is it inherent to the tech?

For the most part no. The transactions are just bigger, you have the send them around to all the full nodes, so you need more bandwidth.

This is not to imply that everyone who wants to use the coin will need a full node though.

[itod]

There's nothing "narrow-minded/resolute/whatever" in being a bit conservative in Bitcoin development. It's not some unimportant altcoin whose failure will bother only a few, it's the core of cryptocoin technology. It's designed to last for a long, long time. If some feature in some altcoin stands the test a time for a year or two, then it should be considered wise to implement it in Bitcoin. IMHO this Gavin's Bloom filter improvement is mindless rush to improve the Bitcoin, it should be tested in some altcoin for some time before upgrading Bitcoin's blockchain processing. If no altcoin developer implements it first, Bitcoin developers should invent an altcoin just to test this feature.

What's wrong with testnet?

At any rate, like I said: "if there is a compelling feature and enough demand for it". I'm not talking about a "chat client" built into Bitcoin Qt. I'm talking about something fundamental like the stealth address rollover. If it is, for instance, revealed that the NSA has provably unmasked 80% of the Bitcoin addresses in use, don't you think that there will be just this sort of clamour? And in that event, if the Bitcoin developers drag their feet, Conformal and Bits of Proof will implement it and release a drop-in replacement for merchant/automation systems, et voila - the blockchain can roll over, existing miners can continue mining without upgrade/moving (unless they want to), and the world continues without needing Bitcoin Core to take point.

You are correct if "NSA has provably unmasked 80% of the Bitcoin addresses in use". Since there's almost no chance that this is a case, this is probably just the Bitcoin core developers response to constant claims about alleged questionable scalability of Bitcoin. The role of testnet is not to play with complete overhaul of block processing that Gavin is proposing, it's to allow the developers to test their own implementations of the current Bitcoin software. This can turn out to be a fantastic win towards Bitcoin unlimited transaction scalability, or big fuckup on the developers side. Why risk it? What's the obvious gain since the Bitcoin can scale for almost an order of magnitude increase of transaction processing with current software? I'm just saying that altcoin launched by BTC core team would be a reasonable line of action.

[tacotime]

thank you for the explanation .. will cryptonote bandwidth be 1/1 on storage requirement increases per transaction over bitcoin?  (Bandwidth requirements = 10X bitcoins)?

CN coins will likely require about an order of magnitude more bandwidth than Bitcoin, yeah.

Is there no way to theoretically mitigate this issue or is it inherent to the tech?

It's inherent to the tech. You might be able to cut it about 50% with gmaxwell/andytoshi's proposal to remove denominations.

[equipoise]

No that's not right. On average every peer receives and transmits each transaction once. Unless you are some kind of supernode (i.e. not a home connection) you won't be getting anywhere near petabytes. On a home connection (sort of the opposite of a supernode) you will likely just receive the transaction once and not send any appreciable fraction of them to any of your peers because they will likely already have them by the time you do.

Yes there will be some inefficiency here, but not (even close to) 100x.

Right.. but that implies supernode centralization, does it not?

When you are running a full node you'll upload/download mostly the block hashes to the peers and not all the transactions - this is enough for decentralization. If you are running a full node you just need each block downloaded once and if the full nodes are doubling each month then each full node upload is O(blockchain size) a month. There are networks currently, which are orders of magnitudes faster then the HDD I/O bandwidth and the limiting factor is the storage I/O bandwidth and not the network bandwidth.

Edit: My 17$ per month internet connection is limited to about 14 TB download and 10 TB upload per month.

[iCEBREAKER]

Here's a genuine Altcon Observation to discuss:

XCN successfully moved to pruned mode, and its price spiked up ~27% from the deep support at 0.00001375.

XCN cracks two hard nuts at once.  Mini-blockchain is ultimate prunability (so a full node may run on your phone).  And withdrawal limits on the account tree enable non-malleable zero-confirmation transactions.

Like E=MC^2, scarcely two dozen people in the world grasp the magnitude of this technology's implications.  

The shilling must end. This is worse than the HashFast thread, where you enticed people into losing millions of dollars.

Thanks for demonstrating my point; even our Legendary TacoTime fails to grasp, much less mention, the importance of XCN's withdrawal limited account tree enabling non-malleable zero-confirmation transactions. 

Making Altcoin Observations is the topic of this thread.  Unless BitFreak pays me to post about XCN, that's not "shilling."

Remember when you accepted compensation to endorse HashFast and "entice" potential customers?  That has nothing to do with Altcoin Observation, does it?  So be nice and don't indulge in off-topic ad hominem tu quoque drama.  You are so much better than that.   

I understand you may be annoyed by the CryptoNight/CryptoNite naming collision, but don't take it out on me.  Plenty of people, including both of us, lost BTC by pre-ordering ASICs.  High-tech is risky and educated people understand the chance of failure is much greater than success.  For every profitable 1st Batch Avalon type launch, there must be 10 BFL style bagholders.  That's simple, dismal economics.

We'd love to have your expert criticism in the XCN thread.  BitFreak and catia are brilliant and very friendly, so come on over!

https://bitcointalk.org/index.php?topic=713538.0

[smooth]

There are networks currently, which are orders of magnitudes faster then the HDD I/O bandwidth and the limiting factor is the storage I/O bandwidth and not the network bandwidth.

At 45 k transactions per second, avg tx size of 2 kbytes (Gavin's numbers) this is only 90 megabytes/sec. Easy for storage. Of course database overhead, etc. Still pretty easy.

[iCEBREAKER]

Perhaps we should refute those claims specifically. The posters here may interested to hear why XCN is not a major advance and a novel solution (I realise it cannot be , because otherwise there would be more buying interest in the XCN market) to blockchain memory and bloat issues.

There must be a number of other alts addressing those issues more successfully (without the shill peddlers)

There was little "buying interest" in the BTC market for several years.  Ditto for LTC and NMC.   

XCN has had the most successful launch of any fair/open-source coin since Monero.

The hashrate is high enough to keep it safe from attacks and price has recovered ~100% from the weekend bottom.

The market is confirming $50k is too small a market cap and it should be upwards of $100k.

The great leap forward to minichains' ultimate pruning is a success and the store is out of cheap coins!



I'm getting ready for the moonshot by accumulating and am not the only one stacking this Litecoin-killer...