On Sun, 2023-03-19 at 10:11 -0700, Kerim Aydin via agora-official wrote: > The below CFJ is 4017. I assign it to ais523. > > snail gained 1 radiance for being co-author of Proposal 8919.
This CFJ was difficult to work out, due to how obscure the wording of rule 2657 is. The intention of rule 2657 is to create a "CAN" when a particular event occurs. However, the rule doesn't outright say what the event is, but instead specifies the event as "each time a player fulfills a scoring condition" and then lists "being a coauthor of a proposal that takes effect" as a scoring condition. One part of the confusion here is down to the scope of "each time" – the implication is that a player can fulfill the same scoring condition more than once, but it isn't specified what counts as the same condition and what doesn't. One possible reading of the condition would be as "being a coauthor of a proposal that has taken effect" – in such a case, this is something that either is true or it isn't, thus the scoring condition in question could only be met once ever. However, there are strong clues that that reading is incorrect (the use of "takes" rather than "taken", the fact that that reading is easy to specify unambiguously yet wasn't, and plain common sense), and that reading fails a rule 217 test: this is a "text is silent, inconsistent or unclear" situation, rather than a case where the reading is the only one that fits the text of the rules, so it should be rejected on "common sense" and "best interests of the game" grounds. As such, the more sensible reading is to interpret the condition in such a way that "being a coauthor of a proposal that takes effect" is fulfilled when the proposal takes effect (and would, e.g., be met twice if the same proposal somehow took effect twice). So the problem boils down to working out what precise timing the rules mean when they state "when a proposal takes effect". This wording doesn't appear directly in rule 2657, but does in the last paragraph of rule 1006 (which is similar in that both rules are attempting to cause side effects of the adoption of a proposal); I'll thus be concentrating mostly on that paragraph of rule 1006 in this judgement (it's "similar but simpler"). The phrase appears a few times in the rules, and is used somewhat inconsistently. Rules 105 and 106 both specify changes to the ruleset as happening "when a proposal takes effect", i.e. the proposal taking effect is what causes the changes to happen (so if we used a model of events in which things happen "simultaneously but in sequence", the proposal taking effect would precede the rule changes). Rule 106 also has a concept of "when a proposal finishes taking effect", which can only reasonably be interpreted as happening after all the proposal's rule changes. Rule 2141 defines rules as "always taking effect", implying that this can be a continuous process. Meanwhile, rule 1006 has a condition "when a proposal takes effect and creates a new office"; this can in isolation be interpreted either as "a proposal creates a new office while taking effect", as "a proposal finishes taking effect, having created a new office", or as "a proposal a) takes effect and b) creates a new office" (i.e. there are two separate requirements that both have to be met). The method that's most consistent with the wording of these rules, collectively, is that the "taking effect" part of proposal adoption works like this: rule 106 specifies that the proposal takes effect, and then defines specifically what "taking effect" means in this context (i.e. the proposal applies the changes specified in its text, and rule 105 gives it permission to make rule changes when doing so). However, trying to generalise the same reasoning to rule 1006 presents a problem: it's possible to interpret it as trying to modify the effect of the proposal to also install a player in an office, which is consistent with the wording of the rules in question and not inconsistent with the intent behind the rule, but would mean that if a proposal with power above 2 attempted to create an office, it wouldn't install a player by default (because rule 1006 would have insufficient power to modify the way in which the proposal took effect). In general, it seems that doing something *during* the resolution of a proposal should require power 3 (or at least power equal to the power of the proposal), due to rule 2150 part 3 stating that a low-powered entity can't affect a higher-powered instrument's operation, and rule 106 implying that the effects in a proposal happen sequentially but instantaneously. (Consider a hypothetical proposal "Create a new office, the Office of Inanity. If the Office of Inanity is vacant, repeal rule 1769." A reading in which rule 1006 attempts to fill the office in the middle of the proposal's resolution seems to be reasonable and consistent with the text, but would clearly be interfering with the operation of the proposal. The situation where the proposal doesn't care about the timing with which the office is filled isn't so obvious, but seems to be analogous – rule 1006 is still changing what the proposal does, in terms of what offices it causes to become filled, and it can't do that with a lower power.) All this implies that, at least for proposals whose power when they take effect is greater than 2 (fortunately, proposal 8919 had power 3, meaning that I don't have to worry about the case of a low-powered proposal), rules 1006 and 2657 aren't "fast enough" to be able to do things "when a proposal takes effect" but before its effect has been fully applied; the only consistent readings that could cause that to happen would be to a) redefine what "takes effect" means, along the lines of "a rule that causes a proposal to take effect also causes its coauthors to be able to gain radiance", or b) change the way in which the proposal was applied. These are both reasonable readings of the text of the rules, but with those readings, the rules would fail to function correctly due to Power considerations. Reading rules 1006 and 2657 instead as "when a proposal has finished taking effect" would cause the rules to function correctly, with no Power considerations required. The problem there is that it involves reading very similar wording in different rules in different ways, according to what would allow the rule to function – rule 106's "when a proposal takes effect" needs to be read as a definition of what it means for a proposal to take effect in order for the rule to function correctly, whereas rule 1006's "when a proposal takes effect" needs to be read as a sort of trigger that applies once the full effects of the proposal have already been applied. That said, this sort of ambiguity (where the same phrase can be interpreted in two different ways) is not unknown in English, and I can't see any reason why the same ambiguous text /shouldn't/ be interpreted in two different ways if the rule 217 tests produce different results in the two cases. Let's think about what would happen if rule 1006 had precedence over rule 106, and both rules had their text interpreted in the same way. The resulting Rules situation there would be "when a proposal takes effect, if it doesn't create a new office, it applies the changes specified by its text in order; if it does create a new office, it instead installs the proposal's author into the office" (because both rules are trying to define what it means for a proposal to take effect). This is clearly absurd, contrary to both common sense (if it isn't applying the changes in its text, it couldn't create a new office, leading to a paradox) and the best interests of the game (we want the proposal system to be reliable and not randomly fail for proposals that would have side effects). As a consequence, the best conclusion here genuinely does seem to be to interpret the phrase in question as ambiguous, with the ambiguity resolved differently in different rules: It's definitely in the best interests of the game for rule 106 to work correctly (if the definition of "a proposal takes effect" becomes broken in such a way that the proposal doesn't make rule changes, Agora is ossified: in that situation, neither the mechanism in rule 106, or the backup mechanism in rule 2034, would be sufficient to make rules changes because both make them because the former causes proposals to take effect and the latter simulates what would happen if they took effect). In fact, it isn't possible for the definition of "takes effect" to break in a way that would prevent rules 106 and 2034 working, because rule 1698 causes prevents any ruleset or gamestate changes that could break it to never happen in the first place. Rule 106 is fundamental to Agora (something that is reflected in the fact that it was part of Agora's original ruleset, and has never been repealed or renumbered, although there have been plenty of amendments). No rule 217 test can possibly interpret rule 106 in a way that would break it, even if this introduces inconsistencies with other rules; the "best interests of the game" are overriding in this case. (Besides, interpreting rule 106 to do something other than what it obviously does would also defy common sense.) All that said, rule 106 doesn't seem to be ambiguous in any way that would affect rule 106 itself, so a rule 217 tiebreak isn't needed here: looking at rule 106 in isolation, it's clear what happens when a proposal is adopted, and the only ambiguity is as to timing details that would only affect other rules (rather than itself). Meanwhile, the interpretation that makes rule 106 work is highly problematic in the case of rule 1006, and by analogy 2657. Unlike rule 106, which is defining the primary effects of adopting a proposal, these rules are intending to define side effects of adopting a proposal, and are clearly "supposed to" be separate from the proposal adoption process, rather than attempting to define a competing proposal adoption process; again, this is a "best interests of the game" and "common sense" argument. The text is silent about exactly when the effects happen, and the wording ambiguous (meaning that we come down to a rule 217 test here): but "after the proposal has finished taking effect" is the only timing that actually functions within the rules. It's generally in the best interests of the game for the rules to do what they seem to rather than actually doing nothing due to being broken in an astonishing way, in cases where both readings are reasonable and consistent with the rules, and the other rule 217 tests don't obviously apply here. (Of course, one reading of the rules is a much better fit for the wording than another, you go with the literal wording of the rule, as rule 217 tells you to, but that doesn't seem to have happened here.) The conclusion is that rule 2654 should be read in such a way that it isn't trying to a) redefine the process for adopting proposals or b) interfere with the process for adopting proposals. If it does neither of those things, then because it doesn't do its thing before the proposal has taken effect, it won't get a chance to do anything until after the proposal has taken effect. The consequence is that the new version of the rule will be able to see that the proposal that created it has taken effect. This is a very complex situation, though, which would ideally be cleaned up legislatively; I think there are a lot of rule 217 calls to be made here (not with respect to rule 106, which is pretty clear with respect to *what* it does even if not *how*, but with respect to the other rules involved), and those calls are inherently somewhat subjective. (I can't see an argument to decide either way based purely on the wording of the rules.) My initial reaction was similar to the caller's: "both rules 106 and 1006 use the same wording, so surely they have the same timing?", but they seem to need to be interpreted differently in order for them both to function correctly; but rule 217 implies that that this sort of case-by-case interpretation can sometimes be correct, and its tests seem to be in favour of using it here. If anything, the wording in rule 2657 (which is different, and confusing in its own right) is more tilted towards an "after the proposal has finished taking effect" timing than the wording in rule 1006; so if rule 1006 is interpreted as occuring with that timing, rule 2657 should be too. All this means that the radiance grant worked. I judge CFJ 4017 TRUE. (But I have lower confidence than usual that I've got this CFJ judgement correct, because the situation is so complicated.) -- ais523 Judge, CFJ 4017
