On Mon, May 12, 2014 at 8:02 PM, Kerim Aydin <ke...@u.washington.edu> wrote: > The rules once said some Item could be "Decreased by N". I found a > scam way to have N come out negative (it wasn't lower bounded). So > of course I tried to "Decrease" the Item by a negative amount and > get lots of said Item (I think a winning amount).
'Twas me, at least in the instance I remember. Unfortunately, I have just discovered that there seems to be no remaining CFJ database accessible via Google, something that as Arbitor I mean to address immediately, but here are two relevant CFJs for your amusement: Detail: http://zenith.homelinux.net/cotc/viewcase.php?cfj=1813 ============================== CFJ 1813 ============================== The AFO won the game as a result of comex's message with Message-ID <6bf32280711281839g6adde4a4wdf21f00d231c7...@mail.gmail.com>. ======================================================================== Caller: root Judge: G. Judgement: FALSE ======================================================================== History: Called by root: 29 Nov 2007 22:01:26 GMT Assigned to G.: 29 Nov 2007 22:10:14 GMT Judged FALSE by G.: 06 Dec 2007 18:18:43 GMT ======================================================================== Caller's Arguments: comex announced 104 (allegedly -- I haven't counted them myself) identical actions, each attempting to spend 0 VCs in order to decrease the AFO's VVLOP by -1. comex did not specify any colors for the 0 VCs e attempted to spend, but I argue that this does not matter, since the number of distinct colors in a set of 0 VCs must be 0. I argue that "spend 0 VCs" is a non sequitur; no VCs are spent, and so no action is performed. I recall there being a precedent on this topic, but I've been unable to dig it up. I argue that "decrease by -1" is also a non sequitur; there is no way to modify an integer value such that it is decreased and such that the amount it is reduced by is -1. pikhq's announcement was similar, but each action read "decreate" rather than "decrease". I argue that this difference is meaningless per Rule 754(i). ======================================================================== Gratuitous Arguments by root: Oh, and I have one other gratuitous argument that I neglected to include when I initiated the case. Rule 2134/1, which empowers the win, does not refer to VVLOP but rather to "voting limit on ordinary proposals". Since there are three different kinds of "voting limit on ordinary proposals" (VVLOP, EVLOP, BVLOP), it is not clear which of those (or, perhaps, the sum of all three) is to be used in determining the win. ======================================================================== Judge G.'s Arguments: CFJ 1813 judgement: The relevant portion of R2126 is: c) A player may spend N+1 VCs of different colors to decrease another player's VVLOP by N (to a minimum of zero). There is nothing limiting the choice of values for the letter N (or any other) in R2126(c). However, the choice of N must result in an action that isn't otherwise forbidden. For example, if the outcome results the transfer of a negative or fractional asset object, it can't be performed. By choosing an N of -1, there is an attempt to "spend 0 VCs" and "decrease VVLOP by -1". Both of parts of this must be possible. Are they? In the old currency fee-based rules, fees were paid by transfers, and a transfer explicitly had to transfer "one or more" objects (R1598/12). So attempts to transfer 0 weren't transfers, so 0 fees couldn't be paid. This was true at the time of precedents in CFJs 1456-1459. This is no longer true, so these precedents no longer apply. In fact, the later Rule governing fees (R1941/2) explicitly allowed fees to be 0 by allowing any "non-negative" cost. So in terms of "spending 0 VCs", old rules support either interpretation, and current rules are silent. CFJ 1444 precedent (on dependent actions, which are regulated similarly and the only still-relevant precedent) implies this is possible. Since it's not forbidden to spend 0 VCs, it's possible to perform an action in this manner. What about "decreasing by -1"? There is nothing restricting VLOPs to non-negative. There is some implication that they should be real numbers in R2156. There is a slight implication they should be non-negative in R683, but it is not compelling (the N in R683 could be 0 for any VLOP < 0). They are not assets, but rather "parameters" which are implied to have integral values but not restricted. Therefore, there is nothing forbidding specifying operations on VLOP using negative numbers. But the term "decrease" needs some interpretation. Arguments have been supplied in discussion that it is meaningless to "decrease" something by a negative amount. In fact, a googling of a linkage of "decrease" and negative terms only yields one context in which the terms are meaningful together. That occurs where an amount of something is decreased, but that something has a negative property. For example, a "decrease in negative ions" in a solution results in positive increase in charge, but that comes about through removing a positive number of negative ions. Once the charge reach zero (there are no more negative ions), one can't raise the charge by removing more imaginary negative ions; at this point, one must add positive ions. If we followed this model, we might allow players to "decrease the negativity" of VVLOP while the VVLOP itself was in negative territory, but not to increase it into positive territory. The complication is that VVLOP is a "parameter" which is implied to be a number. As a pure mathematical quantity, it can be seen as a sum of an arbitrary set of +1's and -1's, and in fact, there is no reasonable reason not to allow this abstraction, we are encouraged by the Rules to use mathematical abstractions by R754(3), and in fact by defining VVLOP as a "parameter" and not an asset, the legislative intent is biased towards such mathematical abstractions. In this context, it is quite possible to remove ("decrease") a certain number of -1's from the equation, if such a thing is done by a mechanism specifically allowed by the Rules, such as in 2126(c). If VVLOP were defined as a type of asset, this would not be possible. But as a parameter, it may even be considered the sum of a set, so it is possible. Let me state for possible appeals this is a judge's reasonable interpretation where the rules are unclear, and so hopefully should be sustainable even if an argument can be constructed for the other position. root wrote: > Oh, and I have one other gratuitous argument that I neglected to > include when I initiated the case. Rule 2134/1, which empowers the > win, does not refer to VVLOP but rather to "voting limit on ordinary > proposals". Since there are three different kinds of "voting limit on > ordinary proposals" (VVLOP, EVLOP, BVLOP), it is not clear which of > those (or, perhaps, the sum of all three) is to be used in determining > the win. R683 states: Among the otherwise-valid votes on an Agoran decision, only the first N submitted by each entity are valid, where N is the entity's voting limit on that decision. which means that R2134 "voting limit" refers to the EVLOP, not the VVLOP. At the time of this CFJ, EVLOP of the AFO is not such that it allows a win, so the result is FALSE. Note that since the message in question will "lead to" a win when EVLOP is set to VVLOP, the same CFJ statement called after the week ends at the end of December 2, 2007 will yield TRUE (unless the VVLOP in question changes in the meantime). ======================================================================== Gratuitous Arguments by G.: I'll add in passing that if VVLOP was even defined as a "number" or "integer" or something, I'd forbid the action. But defining it as a "parameter", where "parameter" is not rules-defined and very broad in its common and mathematical definitions, opens the door to all sorts of mathematical and semantic abstractions (e.g. sets) being performed on it, only limited by the context of the mechanisms allowing the changes. ======================================================================== Detail: http://zenith.homelinux.net/cotc/viewcase.php?cfj=1826 ============================== CFJ 1826 ============================== The AFO's VVLOP changed as a result of comex's message with Message-ID <6bf32280711281839g6adde4a4wdf21f00d231c7...@mail.gmail.com> ======================================================================== Caller: Murphy Judge: root Judgement: FALSE ======================================================================== History: Called by Murphy: 06 Dec 2007 19:15:52 GMT Assigned to root: 09 Dec 2007 23:23:20 GMT Judged FALSE by root: 20 Dec 2007 05:53:31 GMT ======================================================================== Caller's Arguments: In CFJ 1813, Goethe argued that VVLOP is defined as a "parameter", implicitly treated as a number, but could also be interpreted as a set of numbers (added up whenever the value of VVLOP is queried). In this hypothetical context, "decrease by -1" is interpreted as "remove -1 from the set". However, this only works if the set contains a -1 to be removed. This is sufficiently unworkable to shoot down the implicit application of this interpretation. Even if this interpretation were explicitly stated, the AFO's VVLOP set did not contain enough -1's to remove 104 of them. ======================================================================== Judge root's Arguments: When I initially read Judge Goethe's arguments in CFJ 1813, they seemed very reasonable. Upon closer inspection, the interpretation of VVLOP as a summed multiset is reasonable but not necessary. It is equally reasonable to interpret VVLOP as simply a rational number; both interpretations are consistent with the rules. With that in mind, I would adhere to Goethe's precedent if it aligned with the best interests of the game, but I find that it does not. The initiator of these cases argues that to remove a number of -1 values from the set, there must be at least that number of -1 values in the set to begin with. Considering this argument, the set must be finite, or it would be impossible to sum. Furthermore, Goethe's interpretation provides no way to replenish the -1 values in the set, so they must eventually deplete, at which point this interpretation breaks. Since there is no indication as to the initial number of -1 values in the set, there is no way to know when this condition is reached, leading to a nondeterministic split in game state each time the clause is exercised. This is clearly in opposition to the best interests of the game. Additionally, the rules do not only define ways to increase and decrease VVLOP by integral values; Rule 2126(d) defines a means for multiplying VVLOP by fractional multiples of 0.1 and 0. The multiplication could be performed in three ways, none of which are satisfactory: first, a number of copies of the set equal to the multiplicand could be joined together, but this only works when the multiplicand is an integer. Additionally, upon multiplication by 0, the set would be empty and thus would no longer contain any negative values to be removed under R2126(c). Second, every element of the set could be multiplied individually, but this is also incompatible with the interpretation of removing negative values. In particular, upon multiplication by zero, all the elements in the set would be zero and thus non-negative. Upon multiplication by a fraction, it is unknown at what point the set ceases to have integral values, which again results in nondeterminism, and not just when applying the scam. Third, to multiply by a fraction or zero, we could remove copies of each type of element in proportion. In addition to having the same multiplication by 0 issue as the first method, this would result in slight nondeterministic errors when the elements of the set do not happen to exist in the correct proportions. Finally, Rule 2156 defines circumstances under which VVLOP is specifically set to an integer value. Each week, EVLOP is explicitly "rounded to an integer", making it clearly an integer, and VVLOP is then "set to the same rounded value", not to an equivalent set-based value. To support the set interpretation, VVLOP must at some point be changed back into a set, and it is not clear how this must occur. As stated above, I find that all of these points conflict with the best interests of the game, and so I think it important to deviate from the set interpretation. In contrast, a rational number interpretation results in no such problems. Based upon that, and upon Judge Goethe's prior reasoning if VVLOP is assumed to be a number, I judge CFJs 1826 and 1827 FALSE. ========================================================================