Re: DIS: Re: BUS: [Proposal] A degree of inefficiency

[Kerim Aydin via agora-discussion]

On Thu, Jan 23, 2020 at 12:35 PM Alexis Hunt via agora-discussion
 wrote:
> Adding a withdrawal clause prevents the
> initiator from changing eir mind later, which is IMO generally a good thing
> as it means that people can take withdrawal at face value and not have to
> remember to object anyway.

For the record I think this is worth doing on its own for these
reasons (whether or not it's called "withdrawal of intent").  Some
manner of the intender spiking the intent so it needs a new intent to
re-start is useful.  I was just unsure of the wording (especially
since this is the first time I've thought much about how that new
"announcement referenced in paragraph (1)" wording since it was
enacted).

> - The initiator is eligible to support an action by default.
> - With N Support is changed to require N+1 supporters (to account for the
> initiator).
> - Change Agoran consent *only* by requiring more supporters than objectors,
> in addition to the existing requirement. No change to Agoran consent
> otherwise as fundamentally I think that it's a bit of a weird thing that N
> Agoran consent actually requires N*O+1 supporters, where O is the number of
> objectors. This brings it in line with usual supermajority requirements
> being only N*O.
> - Make it so that withdrawing an intent is equivalent to withdrawing
> support, if any, and objecting to it, if possible.

I agree that the counting of the initiator in support-based actions
has bugged me in the past too, and would welcome a change of this
nature (whether coupled with the other change or not).

Re: DIS: Re: BUS: [Proposal] A degree of inefficiency

[Alexis Hunt via agora-discussion]

On Thu, 23 Jan 2020 at 15:11, Aris Merchant via agora-discussion <
agora-discussion@agoranomic.org> wrote:

> Why would we want this? Our system is explicitly designed to make sure
> that people can't announce intent to do something with support or with
> Agoran consent, and then single-handedly tank the intent. That sort of
> thing was the source of scams (or at least I seem to recall hearing it
> was) and required other people to intend to perform the same action if
> they wanted to be safe. To solve the problem, we allow supporters to
> act on behalf to do something even if the original announcer won't;
> this would break that protection. Furthermore, I see no significant
> benefit to
> doing this.
>
> -Aris


Pure support intents do not require any time delay (except if the Speaker
objects, but that's a very edge case), so there is no real cost there. In
effect, "With N Support" means "if N+1 players agree", except that
currently the initiator is awkwardly positioned as the only person who
cannot withdraw eir support, unless e remembered to add a condition "unless
I withdraw this intent", which would be binding.

For With Notice and pure objection intents, the initiator can refuse to
resolve the intent regardless. Adding a withdrawal clause prevents the
initiator from changing eir mind later, which is IMO generally a good thing
as it means that people can take withdrawal at face value and not have to
remember to object anyway.

Agoran consent is the only place where I think your argument holds, again
unless the author remembered to add a binding condition against withdrawal
but that would provide the notification required for others to make a
parallel announcement.

Here's an alternate proposal: make it so that:

- The initiator is eligible to support an action by default.
- With N Support is changed to require N+1 supporters (to account for the
initiator).
- Change Agoran consent *only* by requiring more supporters than objectors,
in addition to the existing requirement. No change to Agoran consent
otherwise as fundamentally I think that it's a bit of a weird thing that N
Agoran consent actually requires N*O+1 supporters, where O is the number of
objectors. This brings it in line with usual supermajority requirements
being only N*O.
- Make it so that withdrawing an intent is equivalent to withdrawing
support, if any, and objecting to it, if possible.

I'd be open to allowing w/o 2+ objection or with notice intents resolvable
by other players, as a separate matter, but we'd have to figure out how to
protect against withdrawal of 2+ objection intents leading players to
ignore them, as is currently a risk.

-Alexis

DIS: Re: BUS: [Proposal] A degree of inefficiency

[Kerim Aydin via agora-discussion]

On Thu, Jan 23, 2020 at 8:28 AM Alexis Hunt  wrote:
> On Thu, 23 Jan 2020 at 11:20, Kerim Aydin via agora-discussion 
>  wrote:
>>
>> On 1/23/2020 7:49 AM, Alexis Hunt via agora-business wrote:
>> > Proposal: A Degree of Inefficiency (AI=3)
>> > {{{
>> > Amend Rule 2595 (Performing a Dependent Action) by inserting ", and did not
>> > subsequently withdraw, " immediately after "published" in the first
>> > paragraph.
>> > }}}

Another question that occurs to me is whether that addition actually
ENABLES people to "withdraw" announcements of intent.  R2124 allows
some flexibility in due to tying it to the definition of "objector"
and making the mechanism about "publicly posting" objections which
includes by sentence context the concept of "publicly withdrawing"
which is a naturally-doable speech act.  (Although withdrawing Support
is nowhere explicitly covered, other than the implication of
possibility in the last paragraph).  On the other hand, withdrawing
Ballots (R683) requires a CAN to enable.  "Announcements of intent"
seem mid-way between - a person cannot generally "withdraw" arbitrary
announcements (e.g. say "nevermind that Notice of Honour I withdraw
it").

Re: DIS: Re: BUS: [Proposal] A degree of inefficiency

[Kerim Aydin via agora-discussion]

On Thu, Jan 23, 2020 at 12:09 PM Gaelan Steele via agora-discussion
 wrote:
>
> I think my (misleading named) “too intense” scam from February of this year 
> might be relevant.
>
> The gist of that was that the rules at the time required that declarations of 
> intent be conspicuous and, *independently*, that intents had to be announced 
> four days before performing the action. Turns out, I could announce the 
> intent inconspicuously, then, four days later, announce *the same intent* 
> very conspicuously and perform the action in the same message.

It's worth noting that the direct result of ths scam was Proposal
8165, which is what added all of the "announcement referenced in
paragraph (1)" language.  So whatever precedents were formed around
that scam (CFJs 3709-10) were legislatively changed (and the new
language is relatively untested compared to our long history of
adjudicating the old language).

> By the way, G: probably because of the renumbering, those CFJs are missing in 
> your archive.

Thanks, yes.  I'll honestly say I keep avoiding that little chunk of
the archives because the CFJ actions around that time are confusing
(not just the renumbering, there's confusing stuff in other adjacent
cases), so I keep looking at it before putting it off and filling in
other archive gaps instead.  :)

DIS: Re: BUS: [Proposal] A degree of inefficiency

[Aris Merchant via agora-discussion]

Why would we want this? Our system is explicitly designed to make sure
that people can't announce intent to do something with support or with
Agoran consent, and then single-handedly tank the intent. That sort of
thing was the source of scams (or at least I seem to recall hearing it
was) and required other people to intend to perform the same action if
they wanted to be safe. To solve the problem, we allow supporters to
act on behalf to do something even if the original announcer won't;
this would break that protection. Furthermore, I see no significant benefit to
doing this.


-Aris

On Thu, Jan 23, 2020 at 7:49 AM Alexis Hunt via agora-business
 wrote:
>
> Proposal: A Degree of Inefficiency (AI=3)
> {{{
> Amend Rule 2595 (Performing a Dependent Action) by inserting ", and did not
> subsequently withdraw, " immediately after "published" in the first
> paragraph.
> }}}
>
> -Alexis

Re: DIS: Re: BUS: [Proposal] A degree of inefficiency

[Gaelan Steele via agora-discussion]

I think my (misleading named) “too intense” scam from February of this year 
might be relevant.

The gist of that was that the rules at the time required that declarations of 
intent be conspicuous and, *independently*, that intents had to be announced 
four days before performing the action. Turns out, I could announce the intent 
inconspicuously, then, four days later, announce *the same intent* very 
conspicuously and perform the action in the same message.

Relevant threads:

Me pulling off and explaining the scam: 
https://mailman.agoranomic.org/cgi-bin/mailman/private/agora-business/2019-February/039934.html
 

DIS thread: 
https://mailman.agoranomic.org/cgi-bin/mailman/private/agora-discussion/2019-February/053227.html

Aris’ judgements for CFJs 3709-10 (originally numbered 3705-6; link is to the 
original judgment, which was then discussed, reconsidered, and reversed in the 
same thread): 
https://mailman.agoranomic.org/cgi-bin/mailman/private/agora-business/2019-February/039940.html
 

DIS thread: 
https://mailman.agoranomic.org/cgi-bin/mailman/private/agora-discussion/2019-February/053244.html
The renumbering: 
https://mailman.agoranomic.org/cgi-bin/mailman/private/agora-business/2019-February/039998.html
 


By the way, G: probably because of the renumbering, those CFJs are missing in 
your archive.

Gaelan

> On Jan 23, 2020, at 9:22 AM, Kerim Aydin via agora-discussion 
>  wrote:
> 
> 
> On 1/23/2020 8:28 AM, Alexis Hunt wrote:
>> More debatable whether:
>> 
>> 1. intent
>> 2. intent again
>> 3. withdraw one intent but not the other
>> 
>> works, but since it refers to "an announcement of intent", the intended
>> interpretation is that it applies to the specific announcement, reinforced
>> by the fact that the other clauses in the rule refer to the specific
>> announcement in point 1; the announcements are clearly not fungible.
> 
> Another thought experiment:
> 1.  intent
> 2.  intent again
> 3.  action that's very direct and explicit in citing intent #1 (e.g. "having
> posted intent in the quoted message (#1), I do X.")
> 4.  It turns out something minor and technical was wrong with the first intent
> that wasn't wrong with the second intent.
> 
> Does it work?
> 

Re: DIS: Re: BUS: [Proposal] A degree of inefficiency

[Alexis Hunt via agora-discussion]

On Thu, 23 Jan 2020 at 12:24, Kerim Aydin via agora-discussion <
agora-discussion@agoranomic.org> wrote:

> Another thought experiment:
> 1.  intent
> 2.  intent again
> 3.  action that's very direct and explicit in citing intent #1 (e.g.
> "having
> posted intent in the quoted message (#1), I do X.")
> 4.  It turns out something minor and technical was wrong with the first
> intent
> that wasn't wrong with the second intent.
>
> Does it work?
>

I believe it does, but it's orthogonal to the proposal, correct?

Alexis

Re: DIS: Re: BUS: [Proposal] A degree of inefficiency

[Kerim Aydin via agora-discussion]

On 1/23/2020 8:28 AM, Alexis Hunt wrote:
> More debatable whether:
> 
> 1. intent
> 2. intent again
> 3. withdraw one intent but not the other
> 
> works, but since it refers to "an announcement of intent", the intended
> interpretation is that it applies to the specific announcement, reinforced
> by the fact that the other clauses in the rule refer to the specific
> announcement in point 1; the announcements are clearly not fungible.

Another thought experiment:
1.  intent
2.  intent again
3.  action that's very direct and explicit in citing intent #1 (e.g. "having
posted intent in the quoted message (#1), I do X.")
4.  It turns out something minor and technical was wrong with the first intent
that wasn't wrong with the second intent.

Does it work?

Re: DIS: Re: BUS: [Proposal] A degree of inefficiency

[Alexis Hunt via agora-discussion]

On Thu, 23 Jan 2020 at 12:12, Kerim Aydin via agora-discussion <
agora-discussion@agoranomic.org> wrote:

> Sorry, I mis-typed, but it doesn't change it IMO.  The question "did I
> publish
> and then subsequently withdraw an announcement of intent" becomes true
> when I
> do it once, and remains true regardless of later announcements of intent.
>

The requirement is that you publish and not subsequently withdraw. That's
not the same as never having published and then withdrawn. Unambiguously,
if I intend, withdraw, and then intend, I have intended without
subsequently withdrawing.


> > More debatable whether:
> >
> > 1. intent
> > 2. intent again
> > 3. withdraw one intent but not the other
> >
> > works, but since it refers to "an announcement of intent", the intended
> > interpretation is that it applies to the specific announcement,
> reinforced
> > by the fact that the other clauses in the rule refer to the specific
> > announcement in point 1; the announcements are clearly not fungible.
> >
>
> I disagree.  If you announce intent twice, then perform the action once,
> there's no real way to say "this action is associated with intent #1 not
> intent #2".  It's really "if at least one of the intents is good, it
> works."
> Similarly, if at least one of the announcements has been "subsequently
> withdrawn", your added language seems to block it.
>
> This is important for how we've done things in the past, especially
> convergences.  If there's an uncertainty in one Intent, publish another
> one.
> If we had 1-1 matching, we couldn't really converge this way without doing
> more specific 1-1 matching in the action announcement, which is a pain and
> prone to error.
>
> -G.
>

If any intent is good, it works, but the conditions on each intent's
goodness are evaluated separately for that intent. For instance, the
conditions in the intent must be specified, and the intent must be within
the time window. You cannot combine intents, by having one intent within
the time window with unsatisfied conditions, and one without conditions but
outside the time window, in order to do the action.

-Alexis

Re: DIS: Re: BUS: [Proposal] A degree of inefficiency

[Kerim Aydin via agora-discussion]

On 1/23/2020 8:28 AM, Alexis Hunt wrote:
> On Thu, 23 Jan 2020 at 11:20, Kerim Aydin via agora-discussion <
> agora-discussion@agoranomic.org> wrote:
> 
>>
>> On 1/23/2020 7:49 AM, Alexis Hunt via agora-business wrote:
>>> Proposal: A Degree of Inefficiency (AI=3)
>>> {{{
>>> Amend Rule 2595 (Performing a Dependent Action) by inserting ", and did
>> not
>>> subsequently withdraw, " immediately after "published" in the first
>>> paragraph.
>>> }}}
>>>
>>> -Alexis
>>>
>>
>> Consider:
>>
>> 1.  Announcement of intent 1.
>> 2.  Withdrawn.
>> 3.  Second Announcement for the same thing.
>>
>> Now, since I have "previously withdrawn" such an announcement, the blocking
>> condition will still be true, I can't do it.
>>
> 
> It's a good thing I wrote "subsequently withdraw" not "previously
> withdrawn", then.

Sorry, I mis-typed, but it doesn't change it IMO.  The question "did I publish
and then subsequently withdraw an announcement of intent" becomes true when I
do it once, and remains true regardless of later announcements of intent.

> More debatable whether:
>
> 1. intent
> 2. intent again
> 3. withdraw one intent but not the other
>
> works, but since it refers to "an announcement of intent", the intended
> interpretation is that it applies to the specific announcement, reinforced
> by the fact that the other clauses in the rule refer to the specific
> announcement in point 1; the announcements are clearly not fungible.
>

I disagree.  If you announce intent twice, then perform the action once,
there's no real way to say "this action is associated with intent #1 not
intent #2".  It's really "if at least one of the intents is good, it works."
Similarly, if at least one of the announcements has been "subsequently
withdrawn", your added language seems to block it.

This is important for how we've done things in the past, especially
convergences.  If there's an uncertainty in one Intent, publish another one.
If we had 1-1 matching, we couldn't really converge this way without doing
more specific 1-1 matching in the action announcement, which is a pain and
prone to error.

-G.

Re: DIS: Re: BUS: [Proposal] A degree of inefficiency

[Alexis Hunt via agora-discussion]

On Thu, 23 Jan 2020 at 11:20, Kerim Aydin via agora-discussion <
agora-discussion@agoranomic.org> wrote:

>
> On 1/23/2020 7:49 AM, Alexis Hunt via agora-business wrote:
> > Proposal: A Degree of Inefficiency (AI=3)
> > {{{
> > Amend Rule 2595 (Performing a Dependent Action) by inserting ", and did
> not
> > subsequently withdraw, " immediately after "published" in the first
> > paragraph.
> > }}}
> >
> > -Alexis
> >
>
> Consider:
>
> 1.  Announcement of intent 1.
> 2.  Withdrawn.
> 3.  Second Announcement for the same thing.
>
> Now, since I have "previously withdrawn" such an announcement, the blocking
> condition will still be true, I can't do it.
>

It's a good thing I wrote "subsequently withdraw" not "previously
withdrawn", then.

More debatable whether:

1. intent
2. intent again
3. withdraw one intent but not the other

works, but since it refers to "an announcement of intent", the intended
interpretation is that it applies to the specific announcement, reinforced
by the fact that the other clauses in the rule refer to the specific
announcement in point 1; the announcements are clearly not fungible.

DIS: Re: BUS: [Proposal] A degree of inefficiency

[Kerim Aydin via agora-discussion]

On 1/23/2020 7:49 AM, Alexis Hunt via agora-business wrote:
> Proposal: A Degree of Inefficiency (AI=3)
> {{{
> Amend Rule 2595 (Performing a Dependent Action) by inserting ", and did not
> subsequently withdraw, " immediately after "published" in the first
> paragraph.
> }}}
> 
> -Alexis
> 

Consider:

1.  Announcement of intent 1.
2.  Withdrawn.
3.  Second Announcement for the same thing.

Now, since I have "previously withdrawn" such an announcement, the blocking
condition will still be true, I can't do it.