If you really want to get political, Satoshi never envisioned SPV the way it's implemented today. Satoshi described SPV repeatedly as clients which accept fraud proofs allowing them to, just like full nodes, reject any invalid chain. Bitcoin was never designed to follow an invalid chain, regardless of it's hashpower. The reason this doesn't exist is Bitcoin is not currently set up to allow for efficient/small fraud proofs, and getting it right is super, super hard.
On March 24, 2017 2:28:15 AM PDT, "Tobias B." <[email protected]> wrote: >Maybe this is so messy because bitcoin was simply never intended to >follow >a minority chain. Maybe the assumption was that it is more reliable to >measure support from economically incentivized miners than economically > >incentivized developers. >Maybe we shouldn't try to keep a minority chain alive in the first >place >and if it wan'ts to stay alive it has to change it's PoW and become an >altcoin with the coin distribution starting from where bitcoin is at >that >point in time. But maybe I'm too political here. You just want to >protect >users, that's fine. But whatever change Bitcoinj implements, users who >do >not update their software will be in danger when there are two bitcoin >chains that're using the same PoW. For me that's an argument to not >support >that scenario. If core really wants to ignore a majority decision of >80% of >the miners then I think they're obligated to change the PoW to prevent >this >mess. > >On Friday, March 24, 2017 at 12:34:45 AM UTC+1, Manfred Karrer wrote: >> >> No I don't bet on it, just wanted to mention that it is a (very) >possible >> further escalation scenario. If the minority chain gets attacked it >seems >> to me also the most likely survival tactic. But I would not count on >that >> of course. >> >> So far I still have no satisfying solution how to deal with a HF in >> Bitsquare. >> Stopping the trading is hard in Bitsquare as it is decentralized. >> Preventing users to use the wallet and become victim of unintended >replays >> is impossible. >> Whitelist for BTC chain nodes helps only for confirmed txs but not >> unconfirmed. That would not risk loss of funds but the usability in >the >> trade would get screwed. >> >> So lets hope that BU does not escalate and we can spend time on more >> productive stuff. >> But at the end it demonstrates a very serious vulnerability with SPV >mode. >> I hope the work on Bitcoin Core SPV will become a viable alternative >soon. >> I consider midterm to move over to that model in Bitsquare. That >would >> solve the privacy issues with bloom filters as well. >> >> >> Am Donnerstag, 23. März 2017 10:46:51 UTC-5 schrieb Matt Corallo: >>> >>> I'm not sure you can bet on BTC having a PoW change, let alone make >that >>> bet and risk users getting completely screwed if it doesn't happen. >>> >>> On March 23, 2017 2:35:02 AM PDT, "Tobias B." <[email protected]> >wrote: >>>> >>>> So as the minority (core) chain is likely to make a PoW change, >wouldn't >>>> this break SPV clients? I hope in such a case this chain would also >add >>>> replay protection - people would have to install new clients >anyway. >>>> >>>> On Wednesday, March 22, 2017 at 10:14:39 PM UTC+1, Manfred Karrer >wrote: >>>>> >>>>> When you talk about minority chain we should take in consideration >that >>>>> this might change over time. So todays winning chain might be the >loser of >>>>> tomorrow and so on. Might be pretty messy with re-orgs and double >spends... >>>>> Also I would not count with what some people expect that the >minority >>>>> chain will quickly die (or get killed). The ETH devs made that >mistake and >>>>> underestimated ETC a lot. Mining resources and price are highly >dynamic >>>>> factors making predictions much harder. Not to talk about the >emotional and >>>>> political factors... >>>>> Also a PoW change is a very likely element in such a scenario... >>>>> >>>>> >>>>> Am Mittwoch, 22. März 2017 11:18:23 UTC-5 schrieb Andreas >Schildbach: >>>>>> >>>>>> Maybe the recommendation should be: if you want to follow a >minority >>>>>> chain, use the Peer API to connect to a trusted peer under your >>>>>> control. >>>>>> You can then run your desired full node implemention on the >trusted >>>>>> peer >>>>>> and bitcoinj will follow. If the network between bitcoinj and the > >>>>>> trusted peer is not to be trusted, use a VPN to secure that >connection >>>>>> (the Bitcoin protocol itself lacks authentication). >>>>>> >>>>>> That would be far more reliable than a version string whitelist. >In >>>>>> fact, frontending bitcoinj with bitcoind has always been a >>>>>> recommendation for centralized/web services, so if they followed >that >>>>>> recommendation they should not have to change anything. >>>>>> >>>>>> >>>>>> On 03/22/2017 03:34 PM, Tobias B. wrote: >>>>>> > Still maybe it adding an API call to go into "whitelist mode" >and >>>>>> then >>>>>> > add certain node ips would be a compromise. Software building >on >>>>>> > bitcoinj could then offer the user to go into this mode and add > >>>>>> certain >>>>>> > ips there. >>>>>> > >>>>>> > On Wednesday, March 22, 2017 at 3:25:15 PM UTC+1, Tobias B. >wrote: >>>>>> > >>>>>> > A whitelist of nodes for me sounds highly conflicting with >the >>>>>> > decentralized nature of bitcoin. Also what if nodes switch >their >>>>>> > software? Not that unlikely in case they notice that their >>>>>> minority >>>>>> > chain is dying. >>>>>> > >>>>>> > On Wednesday, March 22, 2017 at 1:28:09 PM UTC+1, Jameson >Lopp >>>>>> wrote: >>>>>> > >>>>>> > >>>>>> > >>>>>> > On Wed, Mar 22, 2017 at 1:55 AM, Manfred Karrer >>>>>> > <[email protected]> wrote: >>>>>> > >>>>>> > A Bitcoin core node will not mine a transaction >>>>>> originated >>>>>> > in a >1 MB block (not sure if it would relay it). >>>>>> BitcoinJ >>>>>> > has no validation of that consensus rule, so it >would >>>>>> accept >>>>>> > a tx from a BTU block. I think with a whitelist to >>>>>> connect >>>>>> > to BTC core nodes only (or better run your local >node) >>>>>> you >>>>>> > should be safe against receiving txs from a BTU >block >>>>>> (> >>>>>> > 1MB or built on top of a >1MB block). If BTC core >does >>>>>> not >>>>>> > check if a tx has inputs from a >1 MB block for >>>>>> replaying >>>>>> > then we need to take extra care of 0 confirmation >txs. >>>>>> > Does anyone here know if that is the case? >>>>>> > >>>>>> > >>>>>> > If your software ignores all unconfirmed transactions >then I >>>>>> > could see a slightly stronger argument for going the >>>>>> whitelist >>>>>> > route. My point was that unconfirmed transactions will >get >>>>>> > relayed around the network by nodes on both chain >forks. >>>>>> > >>>>>> > >>>>>> > Yes I agree on the protocol level it will be hard >as the >>>>>> > nodes can easily lie. >>>>>> > >>>>>> > Sure wallet providers should not have to deal with >it. >>>>>> But I >>>>>> > don't see much alternative. As pure wallet it might >be >>>>>> > easier to tell the people just dont use yoru wallet >for >>>>>> a >>>>>> > few days/week. But for Bitsquare as an exchange >that is >>>>>> not >>>>>> > really an option, trading should not get stopped >(and in >>>>>> > case of Bitsquare cannot really). >>>>>> > >>>>>> > Maybe 2015 people have not been so nervous about it > >>>>>> because >>>>>> > it was before ETH demonstrated the risk and damage >of a >>>>>> HF... >>>>>> > >>>>>> > Am Dienstag, 21. März 2017 21:42:07 UTC-5 schrieb >>>>>> Jameson Lopp: >>>>>> > >>>>>> > I don't think replay protection can be added at >the >>>>>> > network level. Peers could lie about their >software >>>>>> > version and/or switch it, plus it's highly >likely >>>>>> that a >>>>>> > chain split won't result in a clean network >>>>>> partition. >>>>>> > Transactions will get relayed around the >network >>>>>> > comprised of nodes on both chains and you can >expect >>>>>> > that Core nodes would still relay transactions >to >>>>>> you >>>>>> > that were originally broadcast by Unlimited >nodes >>>>>> and >>>>>> > vice versa. >>>>>> > >>>>>> > At a more philosophical level, I'm not so sure >that >>>>>> the >>>>>> > onus is on every wallet provider to implement >>>>>> protection >>>>>> > against contentious hard forks. You could have >made >>>>>> > similar arguments that BitcoinJ should have >added >>>>>> replay >>>>>> > protection when support for 8MB blocks was >nearing >>>>>> 50% >>>>>> > hashrate back in 2015. >>>>>> > >>>>>> > On Tue, Mar 21, 2017 at 8:28 PM, Manfred Karrer > >>>>>> > <[email protected]> wrote: >>>>>> > >>>>>> > Btw. of course the whitelist can be set by >the >>>>>> user >>>>>> > himself, so if he runs a full node he can >use >>>>>> that. >>>>>> > But knowing that the majority of users are >not >>>>>> > running their own full node we need >alternative >>>>>> > solutions as well. >>>>>> > >>>>>> > >>>>>> > Am Dienstag, 21. März 2017 19:23:56 UTC-5 >>>>>> schrieb >>>>>> > Manfred Karrer: >>>>>> > >>>>>> > Andreas, how are you planning to >protect >>>>>> users >>>>>> > of the Android wallet agains replay >attacks? >>>>>> > How does a user know if she/he can send >his >>>>>> > wallet coins to a peer who accepts only >BTC >>>>>> or >>>>>> > BTU (e.g. exchange)? Otherwise he might >end >>>>>> up >>>>>> > pretty frustrated to see outgoing >>>>>> transactions >>>>>> > at this wallet but not confirmation of >>>>>> receipt >>>>>> > at the receiver. >>>>>> > >>>>>> > You cannot assume that all your users >are >>>>>> > blindly following the longest PoW chain >as >>>>>> the >>>>>> > only valid BTC chain and ignore the >replay >>>>>> > attack risks. >>>>>> > There might be even legal issues >connected >>>>>> to it >>>>>> > (see ETH HF and losses from replay >attacks), >>>>>> > probably less for a wallet provider >than for >>>>>> a >>>>>> > centralized exchange holding customers >>>>>> funds. >>>>>> > But I am not sure if that would not >fall >>>>>> back to >>>>>> > others as well (due diligence), at >least >>>>>> people >>>>>> > could try to sue... >>>>>> > >>>>>> > One of the easiest solution I see is to > >>>>>> deploy a >>>>>> > whitelist of Bitcoin Core nodes and use > >>>>>> those >>>>>> > for the P2P network connections. >>>>>> > You can give the user the choice to >select >>>>>> > between whitelist Bitcoin Core nodes >(if he >>>>>> > wants BTC), whitelist BU nodes (i he >wants >>>>>> BTU) >>>>>> > or public network (if he thinks all >plays >>>>>> out by >>>>>> > itself and is aware of the included >>>>>> risk/mess). >>>>>> > Of course a whitelist is not great as >well >>>>>> but >>>>>> > that would solve the problem that we >cannot >>>>>> know >>>>>> > in advance the data which helps us to >>>>>> > distinguish between the chains (like >blockID >>>>>> or >>>>>> > certain transactions). That whitelist >only >>>>>> need >>>>>> > to be used as long the HF is messy, >once >>>>>> things >>>>>> > have settled it can be removed or >replaced >>>>>> by >>>>>> > other distinguishing data sources like >>>>>> blockIds. >>>>>> > >>>>>> > If BTU implements a clean mechanism to >make >>>>>> it >>>>>> > easy to distinguish that would be the >best >>>>>> > solution of course but I am not aware >of any >>>>>> > concrete plans for such. >>>>>> > >>>>>> > >>>>>> > Am Mittwoch, 15. März 2017 06:08:55 >UTC-5 >>>>>> > schrieb Andreas Schildbach: >>>>>> > >>>>>> > As long as a fork does not change >the >>>>>> proof >>>>>> > of work rules, bitcoinj >>>>>> > makes no assumptions about forks. >It >>>>>> will >>>>>> > always select the chain with >>>>>> > the most work. >>>>>> > >>>>>> > What do you mean by "requesting an >UTXO" >>>>>> and >>>>>> > what do you want to achieve >>>>>> > by that? >>>>>> > >>>>>> > >>>>>> > On 03/14/2017 06:07 PM, Manfred >Karrer >>>>>> wrote: >>>>>> > > If there would happen really a BU >fork >>>>>> SPV >>>>>> > wallets could get a >>>>>> > > connection to a majority of BU >nodes >>>>>> and >>>>>> > so a different view to the network. > >>>>>> > > Any plans or ideas how to deal >with >>>>>> that? >>>>>> > > >>>>>> > > One idea would be to use a UTXO >which >>>>>> is >>>>>> > known to exist on only 1 chain >>>>>> > > request that and use that as a >check >>>>>> to >>>>>> > see which chain the node is >>>>>> > > operated on. >>>>>> > > If it is not the chain the wallet > >>>>>> supports >>>>>> > the node gets disconnected. >>>>>> > > >>>>>> > > Br, >>>>>> > > Manfred >>>>>> > > >>>>>> > > -- >>>>>> > > You received this message because >you >>>>>> are >>>>>> > subscribed to the Google >>>>>> > > Groups "bitcoinj" group. >>>>>> > > To unsubscribe from this group >and >>>>>> stop >>>>>> > receiving emails from it, send >>>>>> > > an email to >>>>>> [email protected] >>>>>> > > ><mailto:[email protected]>. >>>>>> >>>>>> > > For more options, visit >>>>>> > https://groups.google.com/d/optout >>>>>> > ><https://groups.google.com/d/optout>. >>>>>> > >>>>>> > >>>>>> > -- >>>>>> > You received this message because you are >>>>>> subscribed >>>>>> > to the Google Groups "bitcoinj" group. >>>>>> > To unsubscribe from this group and stop >>>>>> receiving >>>>>> > emails from it, send an email to >>>>>> > [email protected]. >>>>>> > For more options, visit >>>>>> > https://groups.google.com/d/optout >>>>>> > <https://groups.google.com/d/optout>. >>>>>> > >>>>>> > >>>>>> > -- >>>>>> > You received this message because you are >subscribed to >>>>>> the >>>>>> > Google Groups "bitcoinj" group. >>>>>> > To unsubscribe from this group and stop receiving >emails >>>>>> > from it, send an email to >[email protected]. >>>>>> >>>>>> > For more options, visit >>>>>> https://groups.google.com/d/optout >>>>>> > <https://groups.google.com/d/optout>. >>>>>> > >>>>>> > >>>>>> > -- >>>>>> > You received this message because you are subscribed to the >Google >>>>>> > Groups "bitcoinj" group. >>>>>> > To unsubscribe from this group and stop receiving emails from >it, >>>>>> send >>>>>> > an email to [email protected] >>>>>> > <mailto:[email protected]>. >>>>>> > For more options, visit https://groups.google.com/d/optout. >>>>>> >>>>>> >>>>>> > >-- >You received this message because you are subscribed to the Google >Groups "bitcoinj" group. >To unsubscribe from this group and stop receiving emails from it, send >an email to [email protected]. >For more options, visit https://groups.google.com/d/optout. -- You received this message because you are subscribed to the Google Groups "bitcoinj" group. To unsubscribe from this group and stop receiving emails from it, send an email to [email protected]. For more options, visit https://groups.google.com/d/optout.
