Re: [bitcoin-dev] Boost Bitcoin circulation, Million Transactions Per Second with stronger privacy

From: raymo@riseup•net
To: Billy Tetrud <billy.tetrud@gmail•com>
Cc: Bitcoin Protocol Discussion <bitcoin-dev@lists•linuxfoundation.org>
Subject: Re: [bitcoin-dev] Boost Bitcoin circulation, Million Transactions Per Second with stronger privacy
Date: Sat, 03 Jul 2021 01:02:09 -0700	[thread overview]
Message-ID: <e843b5c28690557402b72fcd158dc1c2@riseup.net> (raw)
In-Reply-To: <CAGpPWDYAJE4jh=G2g=KSRuLLucEAyZGAD+r4XMpcmw6nk4+Wbg@mail.gmail.com>

Hi Billy,

> What if it was possible for the creditor to claw back the funds
As far as I know the “claw back” mechanism doesn’t exist in Bitcoin
system, and probably most Bitcoiners won’t be agree on it. 
Even if we want to add claw back to Bitcoin in general, and Sabu in
particular, it would add too complexities and uncertainty to Bitcoin. 
So, it would be better to not touch that part, instead focusing on
reduce the cheating risk by putting some penalty for both issuers,
creditors and miners. 
We already have the penalties for both issuers and creditors. 
It looks the miners still can abuse Sabu, but as I told before the miner
or better say the mining pool must be issuer (to be able to sign the
promised UTXO in cheating way) or must be creditor (in order to have a
copy of GT and not lose his money in favor of a stranger miner. Remember
the fact that creditor will lose 70% of their money in favor of Bitcoin
transaction fee in a typical GT) or collaborate one of them in a
conspiracy. Otherwise, there will be no economic benefit in this attack.

All these 3 cases of the attacks, theoretically could be happened, but
the risk to reward ratio is enough high to hinder potential malevolent
from a practical act.
Even this very small risk of miner attacks (which don’t care the attack
costs, since he is not interested in economic benefit, but he wants to
ruin Sabu), would be resolved by a slightly upgrade in Bitcoin protocol
by applying the BIPxxx “for flagging/unflagging promised UTXOs”. 
I am not in rush to apply this upgrade on Bitcoin protocol, instead I am
actively working in order to realize the Sabu protocol and Gazin wallet.
Later the Sabu community will carry the BIPxxx.

Best

On 2021-07-02 17:57, Billy Tetrud wrote:
> Thanks for the details Raymo. A thought occurred to me. Given the fact
> that miners can abuse this system without penalty, it would be useful
> to be able to fix this. What if it was possible for the creditor to
> claw back the funds even if the cheating transaction was mined instead
> of the guarantee transaction? Let's say there was a way to sign a
> transaction that gives the receiver of that transaction the ability to
> override any other transaction that uses the UTXO? If this were
> possible, the issuer could give the creditor this kind of transaction
> as the guarantee transaction, and in the case a cheat was done, the
> creditor could still use the GT to reallocate that UTXO to themselves.
> 
> Now there are issues with this. First of all, it could give anyone the
> ability to double spend. So it would be prudent to limit this in some
> way. The revocation probably should only be valid for up to 6 blocks,
> such that if the transaction has 6 confirmations, it can no longer be
> reallocated (thus preserving the 6 block finality rule). It could also
> be required that the UTXO be marked as opting into this behavior (so
> receivers would know about the possibility it could get revoked). This
> second requirement would require Sabu issuers to make an on-chain
> transaction to set themselves up as an issuer. 
> 
> Another issue is that this would make it possible for transactions to
> expire. Any claw-back transaction would expire 6 blocks after the
> initial transaction happened. This has been generally avoided in
> bitcoin, but I think the relevant issues are solvable. You can find
> additional discussion of that in this thread [1].
> 
> I would imagine this kind of ability would be pretty controversial,
> but since it can close out the possibility for miners to escape
> punishment, it could make this protocol viable. 
> 
> On Thu, Jul 1, 2021 at 3:15 PM <raymo@riseup•net> wrote:
> 
>> Hi Erik
>>
>> Please correct me if I misunderstood.
>>
>>> email is fully compromised.
>>
>> What I got is:
>> Email is not good because the sender and receiver are compromised.
>> Email is not good because the message content is revealed.
>> I can claim same argue about any other client/server model. Since
>> the
>> server (website) service provider will ask some sort of KYC. And
>> even if
>> the server uses end-to-end encryption, the provider company still
>> can
>> read the packets content.
>> In my model the passive listener only can discover who is
>> communicate to
>> whom and make a graph of connections. Although it is a threat for
>> privacy but the server/client model has this flaw inherently, since
>> provider already knew everything about everyone. In my model at
>> least
>> users can make some fake connections and send some fake emails in
>> order
>> to inject noise to communications.
>> Please note the fact that entire communication between mobile
>> wallets
>> (via emails) are asymmetric PGP encrypted. The PGP keys are
>> controlled
>> by end users unlike ALL pretending secure messengers (e.g whatsApp,
>> signal, zoom,…).
>> If you are worried about the way of exchanging PGP public key, you
>> are
>> right. The most secure way is in-person PGP key exchanging.
>> After that for payments the wallets communicate in pgp encrypted
>> messages and they can transfer Bitcoin address through an PGP
>> encrypted
>> cipher, thus no revealing Bitcoin address to public would occur.
>> Neither
>> the amounts of transactions will be reviled.
>> There for it would be a good practice for shops to put their email
>> and
>> PGP public key on shop website and/or PGP public key servers,
>> instead of
>> putting Bitcoin address on website or using 3rd parties services to
>> hide
>> their Bitcoin payment addresses.
>>
>> If I missed some points about “fully compromised” please write
>> it to me.
>>
>>> public keys / addresses are sent
>> As I told before ALL communication in Sabu are PGP encrypted.
>>
>>> other routing data encrypted with public keys
>>> (not sure how data is routed in sabu)
>>
>> Sabu is not responsible for routing at all. It simply sends emails.
>> Indeed the wallets peer-to-peer network in Sabu is pretty straight
>> forward. Each mobile wallet has one email address as its handler and
>> identifier in mobile-wallets-network. Each mobile can send message
>> to
>> another mobile by knowing its email address and the PGP public key.
>> This information can be prepared in first face-to-face contact of
>> mobile
>> owners, or later (something like signing the other’s public key in
>> web
>> of trust) when a creditor wants to spend his money and transfer it
>> to
>> another creditor. The creditor1 send the signed money transfer
>> request
>> alongside the email and public key of creditor2 all in a PGP
>> encrypted
>> message to issuer.
>>
>>> separate the Sabu protocol from the app... allow others to
>> implement
>>> desktop version, or other versions that use other routing systems
>>
>> Indeed, it is my approach too. As I told before users will decide
>> between an unstoppable, permission less, self-sovereignty and
>> decentralized pure peer-to-peer communication network (with some
>> resolvable privacy issues) or some efficient, privacy-mimic central
>> limited network.
>>
>>> you can allow direct-entry of a BIP-word-representation
>>> of a public key/address to avoid privacy/central system concerns
>> Agree. Actually, I was thinking about an easy mechanism to share
>> your
>> public key like what you suggested here.
>> But what I consider for a “central system concerns” is the
>> ability of
>> communication without dependency to any company.
>> As an example, what can you do if the twitter bans your account?
>> Nothing! Your content and entire connections will be lost.
>> But if you form your friends list in your mobile (or computer) and
>> have
>> their PGP public keys and they have yours, and use email as a dual
>> purpose tool. First as a handler (the tool for finding and to be
>> found
>> in internet) and second as a communication tool.
>> Thus, no one can stop you, ban you or limit you to send/receive
>> transaction to/from anyone.
>> What I am trying to say is using email is far better than account
>> (username) in a limited central service like twitter, Facebook,
>> telegram... or even in future Sabu servers!
>> You have your connections under your control in your phone. You can
>> easily change your email and use a new email or even a new service
>> provider without losing your connections and your control over it.
>> You just sign your new email address and send it to your friends
>> circle
>> and notify them about changes.
>> Of course, email is not good for millions of followers but it is
>> obviously good for managing your payment network of hundreds of
>> people
>> (either issuers or creditors).
>>
>> Best
>> Raymo
>>
>> On 2021-07-01 20:49, Erik Aronesty wrote:
>>> your protocol should always assume the email system is fully
>>> compromised, and only send public information over email:
>>>
>>> - public keys / addresses are sent
>>> - other routing data encrypted with public keys (not sure how data
>> is
>>> routed in sabu)
>>>
>>> your end user should be able to verify public keys  / addresses
>>>
>>> - use QR-codes
>>> - phone calls with users reading BIP words out loud
>>> - other in-person information exchange
>>>
>>> separate the Sabu protocol from the app... allow others to
>> implement
>>> desktop version, or other versions that use other routing systems
>>>
>>> -  you can allow direct-entry of a BIP-word-representation of a
>> public
>>> key/address to avoid privacy/central system concerns
>>>
>>> On Thu, Jul 1, 2021 at 4:20 PM raymo via bitcoin-dev
>>> <bitcoin-dev@lists•linuxfoundation.org> wrote:
>>>>
>>>> Hi Billy,
>>>> Sorry for late reply. Let’s jump in proposal.
>>>>
>>>>> Some more information about the benefits of this approach vs
>> alternatives (mainly lightning)
>>>> The most important different is unlike the lightning, in Sabu no
>> one
>>>> have to open a channel and pay Bitcoin transaction fee,
>> subsequently no
>>>> one has to close channel and pay another Bitcoin transaction fee.
>> It is
>>>> the huge improvement since it drops the overhead cost of
>> transactions.
>>>> So, it will be more convenience to trade under Sabu protocol.
>>>> In Sabu none of parties of a transaction are obliged to block
>> money in
>>>> any kind of smart contract or any other m of n signature accounts
>>>> on-chain, so it provides more privacy.
>>>> Since Sabu protocol is designed to motivate people to circulate
>>>> transactions (AKA debt documents) in Sabu network, if every actor
>> act
>>>> rationally no one will aware how much money transferred from who
>> to
>>>> whom.
>>>> In case of fraudulent activity by issuer, the creditor will send
>>>> Guarantee Transaction (GT) to Bitcoin network in order to
>> recapture the
>>>> part of his credit. So, in this case the transaction is literally
>>>> recorded on bitcoin blockchain.
>>>> There is only one another reason to recording transaction on
>> Bitcoin
>>>> blockchain. Where one creditor eager to pay Bitcoin transaction
>> fee in
>>>> order to aggregate thousands or even millions different small
>> amount
>>>> debt-documents in a single transaction on Bitcoin blockchain.
>>>> despite these two cases, the rest of transactions all occur in
>> the Sabu
>>>> network (supposed to be over 99%). Thus, no footprint no
>> bottleneck and
>>>> no over process.
>>>>
>>>> Another important power point of Sabu is its pure-peer-to-peer
>> network
>>>> architecture. In Sabu the mobile wallets communicating to each
>> other
>>>> directly without any central server. There is no centralization
>> at all.
>>>> As a result, there will be no routing as well.
>>>> Since only issuer and creditors are aware of the content of
>> transaction
>>>> (who pay how much to whom) it is a huge privacy improvement,
>> which
>>>> doesn’t exist in other layer 2 solutions.
>>>>
>>>> About the usability of Sabu, although the protocol based on the
>>>> collaborating 2 different peer-to-peer network and 3 classic
>>>> server/client networks, but the end user (mobile wallet user)
>> doesn’t
>>>> see any of these complexities.
>>>> The end user simply installs the mobile/desktop wallet and add
>> her/his
>>>> friends to his phonebook by adding their email address or
>> scanning their
>>>> email (and/or PGP public key). After that s/he can immediately
>> start to
>>>> send/receive Bitcoin through Sabu network. Entire communications
>> between
>>>> wallets are PGP encrypted.
>>>> Another good point in Sabu design is, the 12 seed words are using
>> for
>>>> both Bitcoin wallet private key and the PGP private key. So, it
>> is the
>>>> key of user wealth and its identity as well. For more details,
>> please
>>>> read my previous answer to Alex Schoof.
>>>> The issuer, by using his UTXOs and selling them to creditors earn
>> money.
>>>> the issuer creates the debt document (transaction) by which
>> promises to
>>>> creditor an amount of satoshi. These debt documents are valid
>> Bitcoin
>>>> transaction. The only difference is these transactions are
>> intended to
>>>> circulate in Sabu protocol instead of sending to Bitcoin
>> blockchain.
>>>> Each transaction is a small money transfer. 40,000 Satoshi as
>> input and
>>>> maximum 20,000 Satoshi as credit and minimum 10,000 Satoshi as
>> Bitcoin
>>>> transaction fee.
>>>> The creditors will use these received transactions as money and
>> will pay
>>>> it in exchange of goods or services. For each transaction the
>> creditor
>>>> pays 10 Satoshi as Sabu-transaction-fee to issuer.
>>>> Sabu is not custodial service and the UXTOs are always under
>> issuer
>>>> control, unless issuer or creditor send the signed transaction to
>>>> Bitcoin network. When the transaction was recorded in Bitcoin
>>>> blockchain, the creditor can spend proper UTXO in Bitcoin
>> network.
>>>> Imagine million people use their UTXOs in Sabu, they are issuer
>> and
>>>> issue/update/cancel million transactions per second. All they
>> need is a
>>>> mobile wallet. On the other hand, every one by knowing an issuer
>> can buy
>>>> some Satoshi (whit absolutely no KYC), even 1 Dollar or less, and
>> spend
>>>> it, this time Alice really can buy caffe by Bitcoin ;)
>>>> The Bar can install the mobile wallet and every day receives
>> thousands
>>>> of debt documents (transactions), each worth maximum 20,000
>> Satoshi in
>>>> exchange of coffee. And every evening aggregates those small
>>>> transactions to one single transaction and send it to Bitcoin
>> network.
>>>>
>>>>
>>>> The security model of Sabu is pretty straight forward.
>>>> Issuer is the owner of UTXO(s) which will be used in
>> transactions. The
>>>> issuer is and will the only person who creates transactions and
>> sign
>>>> them. The transactions are valid transaction which either issuer
>> or
>>>> creditor can send them to Bitcoin network, but they will never
>> send
>>>> these transactions to Bitcoin network, because of the high
>> Bitcoin
>>>> transaction fee for each single transaction.
>>>> Since issuer is the only one who can sign transaction (spend
>> UTXOs),
>>>> there is a risk of issuer cheating. And no one can stop issuer
>> from
>>>> cheating, because these are his UTXOs and he has the proper
>> private
>>>> keys.
>>>> The Sabu solution is Guarantee transaction. It is a valid
>> transaction
>>>> that issuer has to sign it alongside the Main transaction. In GT
>> both
>>>> issuer and creditor cut a part of their output in favor of
>> Bitcoin
>>>> transaction fee.
>>>> We suppose miners always seeking for more profit, thus in a case
>> there
>>>> are 2 or more transaction are spending same UTXO as input, miner
>> will
>>>> choose transaction with highest feeRate. There is no economically
>>>> benefit for issuer to cheat creditors and pay less transaction
>> fee
>>>> simultaneously. So rationally the issuer won’t cheat creditor.
>>>> It was the simplest explanation of Sabu security model.
>>>>
>>>>> I agree with others that using email is probably not
>> appropriate for a protocol like this. I would highly recommend
>> making your protocol transport-agnostic, allowing users of your
>> protocol to use any transport they want.
>>>> Indeed, the protocol is transparent-agnostic, if I insist of
>> email as a
>>>> user identifier and communicating tool is because of the idea of
>>>> reforming part of internet architecture and make it more
>> decentralized.
>>>> The wallet users can choose classic architecture. In this case
>> mobile
>>>> wallets will connect to a central server and communicate through
>> that
>>>> server (pretty much like all existed mobile wallets). While some
>> users
>>>> decide to use a pure peer-to-peer communication. I knew email has
>> some
>>>> privacy issues but as always it is a tradeoff. Users can decide
>> between
>>>> an unstoppable, permission less, self-sovereignty and
>> decentralized pure
>>>> peer-to-peer communication network (with some resolvable privacy
>> issues)
>>>> or some efficient central limited network.
>>>> Let me know the critics about email. Hopefully this would lead us
>> to
>>>> improve email instead of letting it die. I strongly suggest email
>>>> because it is the ONLY neutral, free “nonproprietary” and
>> open
>>>> protocol/technology for communication in the world that its
>>>> infrastructure is well-established and is accessible all over the
>> glob.
>>>>
>>>> I tried to explain it more, hope was useful. By the way the
>> complete
>>>> explanation is here
>>>>
>>
> https://raymo-49157.medium.com/time-to-boost-bitcoin-circulation-million-transactions-per-second-and-privacy-1eef8568d180
>>>>
>>>>
>>>>
>>>> Regards
>>>> Raymo
>>>>
>>>>
>>>>
>>>> On 2021-06-22 18:20, Billy Tetrud wrote:
>>>>> I would be interested in seeing some more information about the
>>>>> benefits of this approach vs alternatives up front in this
>> write up.
>>>>> Eg how does the security, cost, usability, and privacy compare
>> to the
>>>>> lightning network, which would be the most likely competitor to
>> this
>>>>> idea. It seems clear that there is more counterparty risk here,
>> so it
>>>>> would probably also be very helpful to compare against
>> traditional
>>>>> custodial solutions as well. If you have specific claims on how
>> this
>>>>> system is better than eg lightning in certain contexts, it
>> would be
>>>>> far easier to evaluate the protocol against those claims, and
>> would
>>>>> also be a lot easier for readers to be motivated to read the
>> whole
>>>>> protocol and do a more full analysis.
>>>>>
>>>>> I agree with others that using email is probably not
>> appropriate for a
>>>>> protocol like this. I would highly recommend making your
>> protocol
>>>>> transport-agnostic, allowing users of your protocol to use any
>>>>> transport they want.
>>>>>
>>>>> On Sat, Jun 19, 2021 at 7:00 PM James Hilliard via bitcoin-dev
>>>>> <bitcoin-dev@lists•linuxfoundation.org> wrote:
>>>>>
>>>>>> I think you're making a number of assumptions about mining
>> that are
>>>>>> not accurate.
>>>>>>
>>>>>>> First of all, how much chance in finding next block the
>> corrupted
>>>>>> miners have? One percent of all Bitcoin hash powers? Or
>> maximum 5
>>>>>> percent or 10? The cheaters must come up in dividing that 1.2
>>>>>> Bitcoin between. After all the risk/reward must fit them. They
>> can
>>>>>> not be a big mining pool since there is no benefit, so they
>> will be
>>>>>> small miners with low hash rate. If they solve the puzzle and
>>>>>> broadcast the block, no one in the entire Bitcoin network has
>> block
>>>>>> transactions or seen it before in their mempool!
>>>>>>
>>>>>> You're making the assumption that miners won't build on top of
>> a
>>>>>> block
>>>>>> with transactions they have not seen before or transactions
>> that may
>>>>>> contain double spends of unconfirmed inputs, this is not how
>> mining
>>>>>> works, as long as the block passes the consensus rules
>> effectively
>>>>>> all
>>>>>> miners will mine on top of it by default, this behavior is
>>>>>> fundamental
>>>>>> to how mining currently works and is fairly deeply baked into
>> the
>>>>>> current mining infrastructure.
>>>>>>
>>>>>>> Will they accept this block? In theory it is possible and
>> have
>>>>>> 0.01 percent chance but we can eliminate this small
>> possibilities by
>>>>>> a simple BIP for miners.
>>>>>>
>>>>>> What would this BIP look like? I don't see how this could work
>> in a
>>>>>> decentralized way as you would need another way of reaching
>>>>>> consensus
>>>>>> on what defines a valid block. Right now the chance is nearly
>> 100
>>>>>> percent that a miner will mine on top of the latest valid
>> block,
>>>>>> many
>>>>>> pools(most last I checked) will even mine on the next block
>> before
>>>>>> they validate the latest block fully(ie validationless mining)
>> to
>>>>>> reduce their orphan rates.
>>>>>>
>>>>>>> We suppose the miners always control transactions with
>>>>>> doc-watchers and avoid accepting transaction with same UTXO
>> but
>>>>>> different output.
>>>>>>
>>>>>> Miners have different mempool policy/rules for what
>> transactions
>>>>>> they
>>>>>> themselves mine but all miners must mine on the most work
>> chain of
>>>>>> valid blocks otherwise they risk their own blocks being
>> orphaned,
>>>>>> any
>>>>>> miner that does not do this is effectively guaranteed to have
>> their
>>>>>> block orphaned right now.
>>>>>>
>>>>>>> Because of high Bitcoin transaction fee, this guarantee
>>>>>> transaction will take place in next block, even if other
>> transaction
>>>>>> which are using the same UTXO as input existed in mempool.
>>>>>>
>>>>>> When a new transaction is broadcast miners do not immediately
>> start
>>>>>> mining on a block template that includes that transaction, the
>>>>>> template won't even be generated immediately when it enters a
>> miners
>>>>>> mempool in practice, for bandwidth/network efficiency reasons
>> mining
>>>>>> pools batch update the stratum templates/jobs they mine
>> against so
>>>>>> there can be significant latency between the time a
>> transaction is
>>>>>> actually broadcast and hits the miners mempool and the time
>> the
>>>>>> miners
>>>>>> actually switch to mining on top it, these batched updates are
>>>>>> essentially like point in time snapshots of the mempool and
>>>>>> typically
>>>>>> remain valid(as in the pool will accept shares submitted
>> against
>>>>>> that
>>>>>> job as valid) until the bitcoin network finds the next block.
>> I
>>>>>> don't
>>>>>> think these batch updates are done more often than every 30
>> seconds
>>>>>> typically, while often it is on the order of multiple minutes
>>>>>> depending on the pool.
>>>>>>
>>>>>> Regards,
>>>>>> James
>>>>>>
>>>>>> On Thu, Jun 17, 2021 at 2:14 PM raymo via bitcoin-dev
>>>>>> <bitcoin-dev@lists•linuxfoundation.org> wrote:
>>>>>>>
>>>>>>> Hi,
>>>>>>> I have a proposal for improve Bitcoin TPS and privacy, here
>> is the
>>>>>> post.
>>>>>>>
>>>>>>
>>>>>
>>
> https://raymo-49157.medium.com/time-to-boost-bitcoin-circulation-million-transactions-per-second-and-privacy-1eef8568d180
>>>>>>> https://bitcointalk.org/index.php?topic=5344020.0
>>>>>>> Can you please read it and share your idea about it.
>>>>>>>
>>>>>>> Cheers
>>>>>>> Raymo
>>>>>>> _______________________________________________
>>>>>>> bitcoin-dev mailing list
>>>>>>> bitcoin-dev@lists•linuxfoundation.org
>>>>>>>
>> https://lists.linuxfoundation.org/mailman/listinfo/bitcoin-dev
>>>>>> _______________________________________________
>>>>>> bitcoin-dev mailing list
>>>>>> bitcoin-dev@lists•linuxfoundation.org
>>>>>> https://lists.linuxfoundation.org/mailman/listinfo/bitcoin-dev
>>>> _______________________________________________
>>>> bitcoin-dev mailing list
>>>> bitcoin-dev@lists•linuxfoundation.org
>>>> https://lists.linuxfoundation.org/mailman/listinfo/bitcoin-dev
>  
> 
> Links:
> ------
> [1] https://lists.linuxfoundation.org/pipermail/bitcoin-dev/2021-June/019050.html