Index ← 3922 CFJ 3923 3924 → text
===============================  CFJ 3923  ===============================

      The quoted message is a distribution message as outlined in
      Regulation AM0.

==========================================================================

Caller:                        Trigon

Judge:                         G.
Judgement:                     FALSE

==========================================================================

History:

Called by Trigon:                                 06 Aug 2021 22:24:15
Assigned to G.:                                   09 Aug 2021 01:44:10
Judged FALSE by G.:                               09 Aug 2021 19:00:38

==========================================================================

Caller's Evidence:

On 06/08/2021 21:38, Trigon via agora-official wrote:
> The Retrieval Period of the Second Victory Auction of Jun 2021 having
> ended at 13:07:41 on 6 August 2021, I hereby bring an end to this
> auction.
>
> ====================
> BIDS ON THIS AUCTION
> ====================
>
> -- FOR THE WIN CARD --
>
> [2021-07-21 05:38:48]     Trigon     400cn
> [2021-07-21 06:12:45]     R. Lee     500cn
> [2021-07-23 00:53:00]     Falsifian  700cn
> [2021-07-23 00:56:16]     Trigon     800cn
> [2021-07-23 23:57:57] *** G.         ( 2^(82,589,933) − 1 )cn
> [2021-07-24 16:49:34] *** G.         (withdrawal)
> [2021-07-27 14:29:49]     Falsifian 1111cn
> [2021-07-30 02:28:28] (3) Trigon    1200cn
> [2021-07-30 05:04:57] (2) R. Lee    1500cn
> [2021-07-30 12:39:17] (1) Falsifian 1717cn
> [2021-07-30 13:07:41] (4) D. Wet      22cn
>
> -- FOR THE JUSTICE CARD --
>
> [2021-07-21 06:12:45] (2) R. Lee       1cn
> [2021-07-23 00:56:16] (1) Trigon      50cn
>
> Bids marked with numbers in parenthesis are active bids.
> Bids marked with asterisks are bids from withdrawn players.
>
> ============
> DISTRIBUTION
> ============
>
> Falsifian wins the first lot for 1717cn.
> Trigon wins the second lot for 50cn.
>
> No other lots exist.
>
> If all of the following succeed I do them:
> { I revoke from Falsifian 1717cn. I grant em one victory card. }
>
> If all of the following succeed I do them:
> { I revoke from Trigon 628cn. I grant em one justice card. }
>


Caller's Arguments: 

This may not be a distribution message because of a textual 
error: I intended to award myself a new justice card for a price 578 
coins greater that what I bid on it. What, if anything, did this message 
do? Did it fail as it did not meet the criteria for a distribution 
message? Did the distribution message succeed and I just violated the 
"SHALL destroy and transfer" bit?

Ambiguous quoted text:

> Trigon wins the second lot for 50cn.
...
> If all of the following succeed I do them:
> { I revoke from Trigon 628cn. I grant em one justice card. }

Relevant part of AM0:

       DISTRIBUTION: The auctioneer for an auction CAN and SHALL, within
       seven days of the ending of that auction's retrieval period,
       create a public message (henceforth the "distribution message")
       that contains a full history of bids on the auction and
       withdrawals from the auction. It must also clearly indicate each
       awardee and the lot e recieves. In this message, the auctioneer
       CAN and SHALL destroy the amount to be paid from the inventory
       each awardee and transfer to that player (or create in eir
       possession if the item is new) the set of assets associated with
       the lot e won. Failing to publish a distribution message
       constitutes the Class 3 Crime of Auction Abandonment.

--------------------------------------------------------------------------

Judge G.'s Arguments:

The interpretations in this judgement are made in keeping with this part
of Rule 2545/5:

>                                     To further aid trade and commerce,
>      auction methods should be interpreted in the name of fairness with
>      deference to the method's clear intent, if intent can be
>      reasonably inferred.

Importantly, fairness etc. is a *requirement* for any auction to happen:

>     e CAN do so by any wholly public method that would be
>     generally recognizable, as specified by the auctioneer at the
>     start of the auction, and under common definitions and terms used
>     in auctions, as a fair, equitable, and timely means of determining
>     the auction winners from among the current players, and enabling
>     the appropriate exchange of goods.

That is, if the method is found to be unfair, inequitable, or untimely, or
doesn't enable the appropriate exchange of goods, the whole auction fails
(because the auctioneer CANNOT conduct an auction in such a manner), even
if that unfair method is written into a regulation.

It is clear, from the presented evidence, that the *intent* of the
regulations used for the initiation and bidding is that the bidding has
determined that:

- Falsifian won the first lot (a new victory card) for 1717cn.
- Trigon won the second lot (a justice card owned by LF&D) for 50cn.
- "Funded" regs ensure that the winners CAN pay for these lots.

Any auction resolution that, after bidding has concluded, allows the fees
to differ from these bids, prevents these exchanges from being made, or
allows a payment or lot award to happen twice, would not be enabling the
"appropriate" exchange of goods.  So if the regulations end up supporting
such incorrect actions (multiple exchanges for the same lot, no exchanges,
exchanges for the incorrect amount), than the interpretation is that no
auction happened at all, because the auctioneer COULD NOT have started it.

Further, in the name of "fairness", it's fairly clear to me that,
following our general Agoran practices for fee-based actions, and also
general practices of debt and consent, that a failed attempt at exchange
(e.g. the auctioneer exchanging the wrong amount or lot) is generally
intended to fail entirely and atomically, like a fee does - even if, in
this case, the auctioneer could make one of the incorrect transfers on eir
own behalf, as authorized by general asset transfer rules.  And further,
to aid trade and commerce, that such a failure would be correctable (i.e.
it wouldn't result in the auction being unresolvable).

One thing that's *unclear* is for multiple lot auctions: if one lot
exchange message is in error, is it "fairer" to have all the lot awards
fail (definitely cleaner overall), or fairer to not hold up the bidders
who got the correct lot for the correct price?  Not sure about regulation
intent and/or fairness there, there's arguments to make on both sides for
what's "fairest".  So that requires a stricter reading of the regulation
text and exact mechanism.

Taking it in parts:

On 8/8/2021 6:44 PM, Telna via agora-official wrote:
>> Relevant part of AM0:
>>
>>      DISTRIBUTION: The auctioneer for an auction CAN and SHALL, within
>>      seven days of the ending of that auction's retrieval period,
>>      create a public message (henceforth the "distribution message")
>>      that contains a full history of bids on the auction and
>>      withdrawals from the auction. It must also clearly indicate each
>>      awardee and the lot e recieves.

This, on its own, gives the full description of what makes a "distribution
message" (I'm interpreting the lowercase 'must' in the last sentence as a
requirement needed to make it a distribution message, not a SHALL).

The regulation makes the "CAN and SHALL" associated with the actual
exchange a second part of the message - but importantly, it's a separate
requirement from the requirement to publish a distribution message, and
separate from the definition of "distribution message":

>>      In this message, the auctioneer
>>      CAN and SHALL destroy the amount to be paid from the inventory
>>      each awardee and transfer to that player (or create in eir
>>      possession if the item is new) the set of assets associated with
>>      the lot e won. Failing to publish a distribution message
>>      constitutes the Class 3 Crime of Auction Abandonment.

So the first thing to notice is that "in this message, the auctioneer CAN
and SHALL" doesn't have a "by announcement".  Now, saying "in a [public]
message, the auctioneer CAN...", when read in light of the special auction
interpretation rules for intent, it's very reasonable to infer that e can
do so by saying so within the message.  However, also importantly, it's
missing some of the by-announcement strictures for how it has to be said,
so some flexibility in expression may be inferred.

Also notice it's a separate CAN and SHALL from the distribution message
CAN and SHALL.   Moreover, note it's a single CAN and SHALL.  "CAN and
SHALL ... each" can be read as being a single action covering everybody.
There's no suggestion that it implies a whole process of actions of the
kind restricted to players "first I transfer the coins of the first
person, then I give them their lot, only after that do I transfer the
coins of the second person..."  And importantly, there's no suggestion
that e violates 4 SHALLs for making a mistake in a 2-lot auction (i.e. one
violation for each half of each exchange).

With this in mind, take the grammatical structure:
>      the auctioneer CAN ... destroy ... and transfer ...

A typical Agoran rules interpretation might find that the CAN is
distributive, to "CAN destroy and CAN transfer".  However, allowing one of
those parts to succeed if the other fails means the auction is no longer
an "exchange".  Since "exchange" is a core auction concept, I find that
"CAN destroy and transfer" is an atomic action respect to each awardee,
similar to fees, and that level of atomicity is required for any auction
method to be called an exchange.

But what about the "each awardee" part?  Is it one atomic action per
awardee, or one atomic action for the whole auction?

First of all, that level of atomicity is not fundamental to "exchange" -
the previous system of allowing winners to pay individually at different
times was perfectly fair.  So that's not core or fundamental to auctions.
 So under the current regulations specifically, let's see what happens if
it's treated as per-player and separable, when an error occurs:

First, Falsifian's award would succeed - it's correct and within a
distribution message.  But the auctioneer would be penalized for breaking
the SHALL with respect to Trigon, since that exchange failed.  To fix
that, the auctioneer would have to post a *second* distribution message,
because the transfers have to happen within a distribution message.  But
that second distribution would leave out Falsifian by necessity, so the
auctioneer would break a second SHALL setting things right (because every
distribution message has to cover *each* awardee).  Or if it doesn't leave
out Falsifian, that means Falsifian might get awarded twice - which
completely violates fair exchanges.  Or the auctioneer could even say "hey
I posted a distribution message and took a penalty for getting it wrong
the first time - I don't actually have to make the second award at all,
there's no remaining SHALL that says I need to".

That's a big mess, and leaves several ways that "fairness" could be
broken, if interpreted that way.  Thus interpreting this particular text
as enabling separable per-awardee exchanges might invalidate the fairness
doctrine of the auction as a whole, any time errors occur.  So I find that
performing the full set of exchanges within a distribution message is a
single, atomic action (with the exchanges being simultaneous, if it
matters).

As an atomic action, it could just as easily be accomplished like this,
without writing out any steps:

> Falsifian wins the first lot (a new victory card) for 1717cn.
> Trigon wins the second lot (a justice card owned by LF&D) for 50cn.
>
> This message serves to make the appropriate exchanges indicated above.

Further,  we have interpretative flexibility.  Even if the auctioneer
lists out individual steps, in a way that implies a sequence, it's still
an atomic action resulting from a single message (if e goes out of eir way
to make it happen in steps - e.g. inserts unrelated actions between the
parts of the transfer, or leaving out any notion of transfer whatsoever -
it departs from the actual CAN allowed to em and just fails).

However, if the transfer steps are specified in greater detail, but one of
the elements is *incorrect*, it would have a different effect.  To see
why, here is part of the message itself:

> Falsifian wins the first lot for 1717cn.
> Trigon wins the second lot for 50cn.
>
> No other lots exist.
>
> If all of the following succeed I do them:
> { I revoke from Falsifian 1717cn. I grant em one victory card. }
>
> If all of the following succeed I do them:
> { I revoke from Trigon 628cn. I grant em one justice card. }

The actual message is deficient in two respects.  First, it needs to
distinguish between new lots, and lots belonging to others (the LF&D).
The auction initiation contained this, but the resolution message did not.
 Second, for Trigon, there is a contradiction between the bid result
(50cn) and the paid amount (628cn).

Not specifying the lot source (LF&D versus new) fails to identify each
lot.  Further, the fact that the price is listed as two amounts throws the
whole distribution message into question.  Even though a first inspection
tells us "oh just the final transfer price is wrong" we could just as
easily think "Trigon forgot a bid of 628 coins in the upper history part,
so it's the 628 that's correct".  We're not directly told what type of
auction this is in this particular message, so an error in one lot may
propagate to other lots, etc.

So, this message, as a whole, did not "clearly indicate each awardee and
the lot e receives".  Rather than being a distribution message that's
missing some parts, it was missing critical information, so wasn't a
distribution message at all.

FALSE.

Some final point to make:

- If the message isn't a distribution
message, it would be possible to read the "I revoke from Trigon 628 coins"
as a typical by-announcement thing that Trigon CAN do as a player.
However, by-announcement actions require setting forth intent to perform
that action.  As discussed above, the auction exchange is a single
atomic act that differs from the by-announcement action, so the intent
expressed was not an intent to transfer coins by announcement.

- This interpretation does allow for multiple "distribution messages" for
a single auction.  An important part of the "intent" interpretation that a
wholly correct distribution message ends the auction, so no further awards
are possible.  This isn't actually stated in the current regs, but is well
within the common definition of an auction and the notion of an exchange -
if it isn't inferable for a particular auction method, the "fair
exchange" part of that auction is broken and the auction would wholly fail
to be initiated.

- This CFJ deals with what happens if an auctioneer's distribution
message is wrong.  There's a different point of failure though - if a
winner puts all eir coins in a contract between the end of the retrieval
period and the auctioneer's otherwise-correct distribution message, that
seems like a "different kind" of failure that might be per-winner, not
per-auction.  And for this kind of failure, it would be "appropriate" as
to the exchange of goods for it to fail while the other transfers succeed.
 But overall I dunno, that's a cfj for another day.

==========================================================================