Date: Tue, 09 Jul 2019 10:31:12 +0000
To: bitcoin-dev <bitcoin-dev@lists.linuxfoundation.org>
From: ZmnSCPxj <ZmnSCPxj@protonmail.com>
Reply-To: ZmnSCPxj <ZmnSCPxj@protonmail.com>
Message-ID: <gq9nEV2JWJaezWYf1OL_vXbqxrhmAKECd4fJ6vCHnXs-VAzK7fNIEJ5Ezd6LhzrxjZA3BdgWC3xZAEJY7ebNQ-QBF6URu29IdAXYgUOZhCM=@protonmail.com>
In-Reply-To: <501EFBBA-8A14-4B64-BD77-1ED5119154EA@gmail.com>
References: <0DBC0DEA-C999-4AEE-B2E1-D5337ECD9405@gmail.com>
	<B853EDF2-8A8A-44B0-A66E-F86175E61EDA@voskuil.org>
	<4mT6iC4Va7Afg15a5NLbddAnF2a_vAcQSXYr_jg_5IyEK2ezblJff7EJZakoqvp4BJlLitt9Zlq1_l5JadR0nVss7VDPW-pv8jXGh7lkFC4=@protonmail.com>
	<A1ADD0BB-F62F-47AF-B043-53BDF3A88CC3@voskuil.org>
	<UyUISFwgh_-KtxpCJonltkqTvVbI9-NBukizE8tKSjB2V12otZiCWQ64sn8oqYk5NDftNHxW3koT9EPOWwVrOkXTP8Dqc-0W0wPGRK-wT34=@protonmail.com>
	<0851B842-34A1-427F-95DC-A1D6AB416FB9@voskuil.org>
	<8D68DC86-1173-43AC-BC84-FE2834741C13@gmail.com>
	<B23C4FC6-2991-4C1B-8B8E-AAA06E9E578F@voskuil.org>
	<501EFBBA-8A14-4B64-BD77-1ED5119154EA@gmail.com>
Feedback-ID: el4j0RWPRERue64lIQeq9Y2FP-mdB86tFqjmrJyEPR9VAtMovPEo9tvgA0CrTsSHJeeyPXqnoAu6DN-R04uJUg==:Ext:ProtonMail
MIME-Version: 1.0
Content-Type: text/plain; charset=UTF-8
Content-Transfer-Encoding: quoted-printable
Subject: Re: [bitcoin-dev] Generalized covenants with taproot enable
	riskless or risky lending,
	prevent credit inflation through fractional reserve
Precedence: list

Good morning all,

I will attempt to restart my thinking from initial principles regarding my =
proposed "Bitcoin Classified Ads Network".

Nodes behave this way:

* Nodes in this network gossip advertisements.
* These advertisements refer to a UTXO that must be unspent at the chain ti=
p considered by each node, else they would be rejected.
* The referred UTXO must contain a commitment to the text of the advertisem=
ent, else the advertisement is rejected.
* Nodes have a maximum limit on the total size of all advertisements they r=
etain and propagate to new nodes, or gossip to their peers.
  This is a deliberate design decision.
* If nodes exceed the above limit, they will sort advertisements according =
to a value-rate, the value of the UTXO divided by the storage size of the a=
dvertisement, and prune advertisements with low value-rate until they are w=
ithin the limit again.
* Once the backing UTXO is spent, the advertisement is removed by nodes tha=
t follow that chaintip.
* As the name ***Classified Ads*** suggests, each advertisement also indica=
tes a "class" in which they belong to.

Then, from the above, we derive how a seller might behave.

* Sellers will attempt to put the minimum possible value into a UTXO commit=
ting to an advertisement, to reduce the opportunity cost of using the value=
 elsewhere.
* Thus the rent of the advertisement in this case is paid to joinmarket mak=
ers and LN forwarding nodes, as the value used in a UTXO backing an adverti=
sement is not useable in joinmarket/LN.
* Sellers remain in full control of their advertising UTXO, and can spend i=
t at any time.
* Sellers may spend part of the UTXO and put the remaining funds into a cha=
nge address that is a new advertising UTXO, and re-transmit the advertiseme=
nt, this time pointing to the new change UTXO.
* However, if the remaining change becomes too low, then its value-rate may=
 drop below the lowest value-rate that BCAN nodes will retain in their (del=
iberately limited) storage, thus also deleting their advertisement from the=
 BCAN.
* Presumably, the reason for advertising at all, is that the seller conside=
rs the cost of advertising to be less than the expected gain of actually se=
lling their product.
* Thus, even if the seller has the ability to spend the UTXO at any time, t=
hey run the risk of spending too much and thus removing their advertisement=
 from the BCAN, and losing the expected gain of having the advertisement ex=
ist on the BCAN.
* A utility-maximizing seller would therefore not spend a minimal-value UTX=
O backing the advertisement until it has gained the advantage of actually s=
elling the product, even if it has the option to do so: it is a forced move=
.
* The cost of keeping the minimal-value UTXO unspent is the opportunity cos=
t in that the value may have been used in joinmarket or LN instead.
* The minimum value will largely be dependent on how much the BCAN is used;=
 more sellers advertising over BCAN will increase the minimum value.
* If the minimum value that is viable to keep its advertisement alive goes =
higher, then the opportunity cost of the seller using the same value elsewh=
ere might exceed the expected gain from selling the product.
  However, this is expected of *any* advertising scheme: if the gains from =
selling is too small to justify the advertisement price, advertising does n=
ot happen; this is expected utility-maximizing behavior.
* If competitors of the seller exist and the BCAN node storage is already f=
illed, competitors can increase the minimum value of a UTXO that can keep a=
n advertisement alive on BCAN by simply adding more of their advertisements=
 to BCAN.
* Thus we expect that, once the BCAN node storage is at or near the maximum=
 value, the minimum value of a UTXO that can back an advertisement will app=
roach the expected gain from selling the product.

Thus the system of simply committing UTXOs to particular advertisement text=
s seems sufficient to extract value from a seller wishing to advertise.
The purpose of this extraction of value is to ensure that spam does not ove=
rload the BCAN.

Let us now consider some kind of specialization, where a HODLer specializes=
 in owning UTXOs, while an advertiser specializes in trading products that =
need advertising of some kind.

* We assume that the specialization means that the HODLer cannot feasibly m=
ake and sell products on its own, while the advertiser cannot own and contr=
ol UTXOs of the minimum value needed to keep their advertisement alive on t=
he BCAN.
* We assume that the specialization means that the advertiser can make and =
sell products for cheaper than the HODLer can, while the HODLer can own and=
 control (and secure) UTXOs of the minimum value needed for advertisements =
to be kept alive, for cheaper than the advertiser can.

Then:

* A HODLer may offer to provide a UTXO locked by a 2-of-2 with a commitment=
 to an advertisement of the advertiser's choosing, in exchange for rent of =
the value, plus an unbreakable promise to return the rented UTXO value back=
 to the HODLer (represented by a `nLockTime` pre-signed transaction that re=
turns the 2-of-2 back to the HODLer control).
* The HODLer is effectively lending the UTXO out to the advertiser, for the=
 time frame agreed upon by the advertiser.
* The rent at which the HODLer lends out the UTXO must be between the oppor=
tunity cost of instead securely utilizing the UTXO in LN or joinmarket, and=
 the expected gain the advertiser expects from having its product advertise=
d.
  * The HODLer is assumed to have the ability to secure the UTXO and retain=
 all data it needs to recover the UTXO; this is part of the assumption that=
 the HODLer specializes in such.
  * The advertiser is assumed to have positive gains from creating, adverti=
sing, and selling its product; this is part of the assumption that the adve=
rtiser specializes in such.
* The HODLer and advertiser can agree to refund part of the rent, if the ad=
vertiser signs a transaction that immediately returns control of the value =
to the HODLer, before the agreed `nLockTime`.
* The above constructions can be done in current Bitcoin.
* However, the same constructions could be done with a covenant as proposed=
 by Tamas, possibly with reduced communication/coordination costs between t=
he advertiser and HODLer.

Now, there remains the question as to whether users will actually patronize=
 the BCAN instead of existing advertising systems.

* We assume that privacy is valuable to users.
* We assume that users of BCAN will run BCAN nodes.
  This leaks them as users of BCAN, a small loss of privacy.

Then:

* Users can look for advertisements of specific classes by simply querying =
their own BCAN node.
  This does not leak privacy ata all as long as the communication channel o=
f the user with their own BCAN node is private.
  * Compare this to alternatives, which involve some entity observing the b=
ehavior of users and thus invading their privacy.
* Advertisers that misclassify their advertisements will be unable to reach=
 their target audience.
* Utility-maximizing advertisers will correctly indicate the class of their=
 advertisements, as otherwise they would be paying the advertising cost wit=
hout gaining the benefit of the advertisement.


Regards,
ZmnSCPxj