election by jury (www.electionbyjury.com/manifesto)
-
@cfrank well thank you very much for reading it. I know it is very long but I think anyone looking to seriously consider and criticize the system should read it so I'm glad you did.
there is no perfect information environment but I'm just saying it's better than any other solution that could possibly be created. and how sequestered you are is completely tunable. in principle you could sequester the jury 3 months ahead of the election or a year ahead of the election or two years agrad of the election if you think it's really that important. I'm not particularly taking any stance on that.
-
@cfrank said in election by jury (www.electionbyjury.com/manifesto):
@clay I read through most of your manifesto
Just for the avoidance of doubt, I did also read through most of it, even if I did miss one or two things!
-
@clay the more I think about this, the more I agree. Are there any initiatives for this to be done locally anywhere in the USA?
I do think some of the most sensitive aspects of implementing this kind of proposal would relate to the need for auditability by volunteer citizens or the press, and the possible need for an insulated environment for the jury, and those seem partly in conflict.
I already indicated my thoughts about the sampling rules for the jury in terms of district-based representation for higher level offices. Sometimes I wonder if a completely hierarchical system makes more sense, where districts elect representatives and electors for higher offices, and those electors handle the higher order coordination with other districts, etc.
-
it's done in georgia.
https://www.electionbyjury.org/learn-more/ebj-in-georgia
https://electionbyjury.substack.com/p/the-henry-county-test-what-happensi don't know what you mean by "auditability". it's a simple secret ballot, where all the votes can be cast into a translucent box, and counted right there on the spot in full view of the jury. it's just a "really small election". of course a jury has an insulated environment.
I already indicated my thoughts about the sampling rules for the jury in terms of district-based representation for higher level offices.
i have no idea what you're talking about. the sample would be from the exact same pool of voters as whatever office you're talking about. e.g. for senate, it would be a whole state.
I wonder if a completely hierarchical system makes more sense, where districts elect representatives and electors for higher offices, and those electors handle the higher order coordination with other districts, etc.
i have no idea what you're talking about.
-
@toby-pereira thanks.
-
@clay I’m being pretty clear about what I mean. Please take some time to reciprocate the level of engagement I have given to your manifesto. I’ve explained my concerns in multiple ways; if a particular part remains unclear, you can ask a specific question about it.
By auditability, I mean of the process as a whole, by citizens who are not selected as jurors. The randomness, the information the jury is given, etc. are things that the public deserves the ability to audit for correctness and transparency. I don’t mean the secret ballots of the jurors.
“of course a jury has an insulated environment.”
Yes, that’s my point. An insulated environment is potentially in conflict with demand for external auditability of the process.
In terms of sampling, I am not asking which existing electorate should supply the random sample for an existing office. I understand that, under your proposal, a Senate jury would currently be sampled statewide.
I am questioning the prior institutional choice to treat the state as one undifferentiated constituency for that office, in the sense of sampling uniformly from the entire statewide population without guaranteeing representation to the local political units by which many interests and concerns are naturally organized. An alternative would give districts or localities guaranteed representation in higher-level decision-making, while still using equal random sampling within each locality.
By a hierarchical system, I mean one in which local constituencies select their own representatives or electors, and those delegates coordinate at progressively higher federal levels, rather than every higher-level institution drawing directly from the full aggregate population. In an election-by-jury context, representatives of localities could even negotiate terms under which their constituencies might be represented in higher-level juries—for example, the allocation of juror seats, guaranteed minimum representation, and the territorial units from which jurors are sampled.
These are questions about constituency structure, federal representation, and delegated authority. They also aren’t proposals for unequal sampling among individuals within a fixed constituency. They’re legitimate institutional questions that others evaluating the proposal are also likely to raise.
I also looked more closely at the Georgia examples, and to me they reinforce that these questions are not merely hypothetical. Georgia’s grand-jury appointment systems explicitly specify territorial representation. For example, the Cherokee County Water and Sewerage Authority has six directors appointed by the county grand jury: four representing separate districts, one representing the county at large, and one representing a participating municipality. The elected chair of the county commission serves as the seventh director:
https://ccwsa.com/about__trashed/history-of-ccwsa/
Henry County’s ethics-board law likewise required its five grand-jury appointees to reside in different commission districts, while Crisp County combines four grand-jury-appointed power commissioners with three elected county commissioners.
It would be a substantial research project to look into these implementations in detail, but my narrower point is that actual grand-jury appointment systems have incorporated additional political choices in line with the topics I’m trying to discuss.
-
@cfrank said in election by jury (www.electionbyjury.com/manifesto):
@clay I’m being pretty clear about what I mean.
it certainly was not clear to me.
Please take some time to reciprocate the level of engagement I have given to your manifesto. I’ve explained my concerns in multiple ways; if a particular part remains unclear, you can ask a specific question about it.
the manifesto is voluminous, specifically to avoid ambiguity. a few sentences from you in this forum are not voluminous, and are unclear to me. i did "ask a question", in that i called out the specific excerpt that i found incomprehensible.
By auditability, I mean of the process as a whole, by citizens who are not selected as jurors.
more ambiguity.
The randomness, the information the jury is given, etc. are things that the public deserves the ability to audit for correctness and transparency.
okay, finally some specifics. altho it still strikes me as fundamentally confused. it goes without saying that you want a provably random process. we already have this in our jury selection system, so there's nothing novel here. we have randomness for lotteries to. it goes without saying we want a provably random process.
as for "information the jury is given", that does not need to be "audited" whatsoever. tho i see no particular reason not to publish the entire transcript, it's not particularly relevant. the jurors themselves "audit" it, in that they are there witnessing the candidates make their arguments. every single candidate observes every other candidate's complete arguments. what would "auditing" by the public even mean? what is "correctness"? if a candidates makes false claims, it's the other candidates' prerogative to point that out in their rebuttals. that's the process. there's nothing for the public to "audit". you don't get to litigate it and go back and tell the jury, "no, that was a lie." that's the other candidates' job! if you insert a judge into the process, who gets to determine what is fact or not, then you've defeated the whole point, of ensuring neutrality thru omni-partisan rhetoric. you allow bias.
the manifesto painstakingly explains all of this.
“of course a jury has an insulated environment.”
Yes, that’s my point. An insulated environment is potentially in conflict with demand for external auditability of the process.
-
you haven't shown how we need auditability in terms of what goes on in the courtroom, beyond the candidates themselves being able to surface if any part of the process was violated.
-
you seem to not be considering that insulation has two separate directions. we want to prevent external information from entering the courtroom. no one ever said we need to insulate the public from receiving information from the process.
I am questioning the prior institutional choice to treat the state as one undifferentiated constituency for that office, in the sense of sampling uniformly from the entire statewide population without guaranteeing representation to the local political units by which many interests and concerns are naturally organized.
you want uniform sampling, so you get statistical validity. you want to avoid biasing the process by having arbitrary guaranteed representation (read, overrepresentation) of any arbitrary political interest. the point is to use a sufficiently large sample size that the interests of any non-negligibly-sized group shows up already. then if you get to the point where the interests of one person in a million aren't registered, that's fine, because it's statistical noise. "bob would have moved the election result a millionth of a percent, but we didn't hear his concerns—horrors!"
An alternative would give districts or localities guaranteed representation in higher-level decision-making, while still using equal random sampling within each locality.
that would be extremely bad, as it would add bias and make the result less statistically accurate.
Henry County’s ethics-board law likewise required its five grand-jury appointees to reside in different commission districts, while Crisp County combines four grand-jury-appointed power commissioners with three elected county commissioners.
you're confusing different things. that's not about the jury, it's about the commissioners. and this is basically a messy request for proportional representation. the optimal way to solve for that would be to just use a proportional voting method. but you don't want to bias for specific kinds of proportionality. you just want to let the voters decide. if i would rather have a guy in another district represent me, that's my choice. maybe he's much more aligned with me on a multitude of policy details, and there's very little that hinges upon which district we live in. the district mandate is putting a thumb on the scale rather than letting voters decide.
and even proportional representation would be questionable, because there's no clear evidence it's better.
https://clayshentrup.medium.com/the-proportional-representation-fallacy-553846a383b3It would be a substantial research project to look into these implementations in detail, but my narrower point is that actual grand-jury appointment systems have incorporated additional political choices in line with the topics I’m trying to discuss.
no, this is just confusion between the jury and the elected officials. and even the rules they did put in place are obviously suboptimal for simple logical reasons i just outlined.
-
-
“It certainly wasn’t clear to me.”
You have made that clear. My point is that you have not identified anything incoherent in what I wrote.
I said:
“I do think some of the most sensitive aspects of implementing this kind of proposal would relate to the need for auditability by volunteer citizens or the press, and the possible need for an insulated environment for the jury, and those seem partly in conflict.”
That is a coherent statement. It identifies a possible tension between external scrutiny of the process and insulation of jurors from outside influence. You could reasonably have asked which parts of the process I thought should be auditable, or what form I thought the conflict might take. Instead, you said that you had no idea what I was talking about.
Likewise:
“By auditability, I mean of the process as a whole, by citizens who are not selected as jurors.”
The intended distinction is between scrutiny conducted internally by the jurors and scrutiny conducted externally by members of the public or press. That may require further specification, but it is not incomprehensible.
You now say that the transcript could be published, but that there is nothing for the public to audit because candidates can rebut one another. That does not resolve my concern. Publishing the information presented to the jury would allow the public and press to examine the claims, omissions, procedures, and informational environment that influenced the selection. Whether that scrutiny should create a formal remedy is a separate question from whether the process should be publicly inspectable.
You also say:
“you haven’t shown how we need auditability in terms of what goes on in the courtroom.”
Whether you think it is strictly necessary is not the only relevant question. A public demand for transparency and independent scrutiny would predictably arise around a process that selects public officials. That demand would have to be accommodated, rejected somehow, or limited through explicit institutional rules, which might themselves be unacceptable to the public.
You write:
“we want to prevent external information from entering the courtroom. no one ever said we need to insulate the public from receiving information from the process.”
Then we agree that information should flow outward. The remaining issue is the restriction on information flowing inward. Preventing selected jurors from receiving external political speech, reporting, or commentary would create tension with ordinary public political communication and would be difficult to enforce without intrusive restrictions on jurors’ media exposure and contact with the outside world.
On representation, you write:
“you want uniform sampling, so you get statistical validity.”
No. “Statistical validity” is not meaningful until the target of representation has been specified. Uniform statewide sampling is appropriate if the objective is to approximate the aggregate preferences of individuals in the state. It is not politically neutral to assume that this is the only legitimate object of representation.
My concern is that uniform statewide sampling may fail to give institutional standing to territorially organized interests. Large political systems almost always include geographically stratified representation because locality can carry politically relevant interests that are not reducible to individual population shares.
You then say that guaranteed territorial representation would introduce bias and reduce statistical accuracy. That conclusion follows only if the target has already been defined as the undifferentiated statewide population, which is precisely what I am contesting. If the institution is intended to represent both individuals and constituent localities, territorial stratification is part of the target rather than a distortion of it.
You also write:
“the district mandate is putting a thumb on the scale rather than letting voters decide.”
It is not simply a thumb on the scale. It is a different choice of representative unit. District-based representation guarantees territorial coverage and can reduce the redundancy produced when large numbers of people with correlated local interests are treated only as independent statewide observations. It may also ensure that smaller or geographically concentrated communities are present in the decision-making process.
A federation need not allocate political influence solely by counting individuals within one undifferentiated population. Distinct localities may require institutional standing because their interests, conditions, and vulnerabilities are not captured adequately by raw population weight alone.
That structure can certainly be designed badly. District boundaries can be arbitrary, unequal, or manipulated. But it is not therefore “obviously suboptimal.”
That is the point I have been trying to make. You are treating uniform statewide representation as neutral and alternatives as biased. I am saying that both arrangements embody prior political judgments about what the institution is supposed to represent.
There is also an institutional problem hidden inside the idea of insulation that @Toby-Pereira alluded to. Insulation has to be enforced by some government body or other administrative authority. That body would determine what information jurors may receive, what contact counts as improper, how violations are investigated, and what consequences follow. Those decisions could have a substantial effect on the election itself.
In other words, the institution charged with protecting the jury from outside influence could acquire outsized political influence through control of the jury’s informational environment and selective enforcement of the rules. That authority would itself need strong limits, transparency, independent review, and public auditability.
-
Large political systems almost always include geographically stratified representation because locality can carry politically relevant interests that are not reducible to individual population shares.
this is just statistical nonsense.
first, notice this argument has nothing to do with sampling. whatever force it has applies identically to a full-electorate vote—it's a claim that localities should have standing beyond their population, i.e. that some people's votes should count more by address. and "interests not reducible to individual population shares" doesn't survive inspection: interests are held by persons. if a locality has an interest, the people there have it, and a uniform sample captures them in exact proportion to how many they are. the only content left in "not reducible to population shares" is that the interest should count beyond the people who hold it. that's the whole claim, stated plainly.
as for why large systems historically stratified by geography: partly union-formation bargaining (above), and partly 18th-century logistics. before statistics and telecommunications, geography was the only available technology for making a legislature resemble the country. adams said the representative body should be a portrait of the people in miniature—districting was the 1780s approximation of that portrait. random sampling produces the portrait exactly. a geographically concentrated minority that districting was invented to protect is represented on the jury automatically, in proportion, with no lines drawn and no gerrymander possible. so the historical prevalence of geographic stratification is evidence for sortition: constitution-builders have always wanted descriptive representativeness and had to hack it with maps. the jury retires the hack.
It is not simply a thumb on the scale. It is a different choice of representative unit. District-based representation guarantees territorial coverage and can reduce the redundancy produced when large numbers of people with correlated local interests are treated only as independent statewide observations. It may also ensure that smaller or geographically concentrated communities are present in the decision-making process.
everything you just said here is repeating what I already pointed out. this is a statistical distortion away from actual accurate representation.
A federation need not allocate political influence solely by counting individuals within one undifferentiated population. Distinct localities may require institutional standing because their interests, conditions, and vulnerabilities are not captured adequately by raw population weight alone.
that makes no sense. you could just as well say Brian's vote should count for twice as much because his political preferences are rare.
That structure can certainly be designed badly. District boundaries can be arbitrary, unequal, or manipulated. But it is not therefore “obviously suboptimal.”
anything that is not just a random sample is obviously suboptimal, based on elementary statistics.
That is the point I have been trying to make. You are treating uniform statewide representation as neutral and alternatives as biased.
and that is absolutely correct.
I am saying that both arrangements embody prior political judgments about what the institution is supposed to represent.
this is wrong. there is no "supposed to". there is no "ought". there is just what rational voters would prefer behind the veil of ignorance.
-
Moderator note: @clay, your tone is crossing the line for productive exchange.
Referring to another participant’s argument as “statistical nonsense” and “statistically illiterate nonsense,” and describing that participant as “deeply confused,” is not acceptable here. You may identify errors, criticize reasoning, and reject premises. Do not characterize another participant’s intelligence or competence.
The disagreement concerns the appropriate target and unit of political representation. Treating one side of that contested institutional question as evidence of statistical illiteracy is personal disparagement, not argument. Continue without such characterizations.
Returning to the substance in my capacity as a participant:
Districting is not merely a “hack.” It is a choice to give territorial constituencies standing within a federated political structure. Your argument assumes that the only legitimate target is the distribution of individual preferences across the entire population. That is one possible target, but it is not politically neutral and cannot be established by statistics alone.
Suppose 80 percent of a state’s population lives in a few coastal metropolitan areas and 20 percent lives across a large rural interior. A statewide uniform sample may accurately reproduce the numerical distribution of individual preferences. It does not follow that it provides an adequate mechanism for coordinating the distinct territorial interests, jurisdictions, infrastructure, resource use, and local conditions of the rural interior. A federation may deliberately give constituent territorial units institutional standing rather than permit every issue to be determined solely by aggregate population weight.
You write:
“this is a statistical distortion away from actual accurate representation.”
“Accurate representation” is undefined until you specify what is being represented. If the target is individual preferences in statewide proportion, then uniform sampling is accurate relative to that target. If the target includes representation of constituent territorial units, then stratification is part of the target rather than a distortion, as I have already stated. This seems to be a point of disagreement. If you believe this notion is erroneous, can you please identify the statistical or political premise you reject, and illustrate why?
Politics is not merely an exercise in estimating a population distribution. It is also a system for coordinating interests, jurisdictions, rights, obligations, and policies across differentiated communities. The desired representation must be judged relative to that political purpose. Statistics cannot choose the purpose for us.
You write:
“the district mandate is putting a thumb on the scale rather than letting voters decide.”
No. It selects a different representative unit. I am not arguing that an individual’s vote should count more because that individual’s preferences are rare. I am arguing that a federal structure may reasonably represent both persons and constituent territorial communities in order to coordinate geographically organized interests and prevent persistent statewide majorities concentrated in some regions from overriding sufficient political investment in smaller constituent regions.
Your example involving “Brian’s vote” therefore misses the argument. Territorial representation is not a reward for unusual preferences. It gives institutional standing to communities whose concerns are organized partly by geography: land use, water, transportation, extraction, agriculture, environmental exposure, infrastructure, policing, and local jurisdiction.
Geographic representation can certainly be designed badly. Districts may be unequal, arbitrary, or gerrymandered. But none of that proves that territorial representation is inherently irrational or “obviously suboptimal”; it shows that the structure must be designed carefully. Treating the entire state as a single constituency lies at the opposite end of the same design spectrum and is itself a districting and aggregation choice. For a large, territorially heterogeneous polity, that arrangement is not self-evidently ideal or stable.
You also say:
“there is no ‘supposed to’. there is no ‘ought’.”
That cannot be correct. The decision to define representation as proportional reproduction of individual statewide preferences is itself a normative institutional choice. Invoking what rational citizens would choose behind a veil of ignorance does not eliminate the normative question. A person who did not know whether they would belong to a dense metropolitan majority or a smaller rural region might reasonably prefer a system that protects both individual equality and some degree of territorial standing. They might reject a structure in which a geographically concentrated majority could permanently determine policies for regions with substantially different conditions and needs. The veil of ignorance therefore does not uniquely imply undifferentiated statewide representation.
The proper object of representation cannot be derived from sampling theory alone, because sampling theory evaluates procedures relative to a target that must first be specified. You have assumed a political objective and then labeled alternatives statistically inaccurate relative to that assumed objective. That does not establish that the objective itself is uniquely correct.
If you believe my position contains a contradiction, identify the propositions that generate it. Calling the position statistically illiterate does not demonstrate one.
-
Moderator note: @clay, your tone is crossing the line for productive exchange. Referring to another participant's argument as "statistical nonsense"
tone policing. substitute "statistically nonsensical" or "statistically invalid" or whatever register you prefer. the claim is about the argument, and it stands, so let's test it.
The disagreement concerns the appropriate target and unit of political representation. Treating one side of that contested institutional question as evidence of statistical illiteracy is personal disparagement, not argument.
the boundaries of a constituency are a political given—states, districts, whatever. i've said that from the start. but given a constituency, the claim that its members should be weighted unequally by address is the thing i'm calling invalid, and here's the two-step derivation. behind a veil of ignorance you have an equal chance of being anyone in the constituency, which means you don't know your address. so address-weighting cannot improve your ex-ante prospects—under any risk attitude, from pure expectation to pure worst-case. no symmetric evaluation of prospects assigns you a gain from privileging an address you have equal odds of not holding. harsanyi formalized the expected-utility version; the symmetry step alone is enough here.
Districting is not merely a "hack." It is a choice to give territorial constituencies standing within a federated political structure.
"standing for territorial constituencies" is address-weighting, named politely. it weighs some persons more than others by where they live. nobody behind the veil chooses that, per the above.
Your argument assumes that the only legitimate target is the distribution of individual preferences across the entire population.
no assumption required. the office defines the target: a senator for the state answers to the state's population, so the estimand is what that population would conclude, and the unbiased estimator of a population quantity samples the population as it is. your alternative target—"representation of constituent territorial units"—amounts to a different office, one answering to a federation of localities. that's the boundary question again, which is upstream politics, and which i've never claimed the mechanism settles.
That is one possible target, but it is not politically neutral and cannot be established by statistics alone.
inverted. anonymity—every person counts once—is the unique neutral weighting: it's the only rule that doesn't reference who or where you are. your proposal is the non-neutral one by construction, since it requires knowing which addresses get the multiplier.
Suppose 80 percent of a state's population lives in a few coastal metropolitan areas and 20 percent lives across a large rural interior. […] It does not follow that it provides an adequate mechanism for coordinating the distinct territorial interests, jurisdictions, infrastructure, resource use, and local conditions of the rural interior.
run your own scenario. the harm you're insuring against is intensity mismatch: a numerical majority with weak preferences overriding a minority with strong ones. that's an artifact of ordinal, one-person-one-vote aggregation. cardinal aggregation—which a jury using score voting closely approximates—counts the rural minority's stakes at full weight, issue by issue, because it elicits those stakes from each voter directly. and this is measured, not assumed: strength of preference is exactly what VSE simulations score methods on, with normalized utilities and strategic voters priced in. now suppose you think some residual inaccuracy remains. address-weighting cannot fix it, because a correction requires knowing whose utilities are undercounted and by how much, and you cannot read minds. a geographic multiplier applies a constant to a voter's entire preference vector on zero information about his actual stakes: it overcounts him where he has no special interest, adds nothing where his interest is extreme, and inherits every measurement error the ballot already had, plus an arbitrary distortion on top. an instrument that asks each person his stakes, however imperfectly, dominates one that guesses them from his zip code.
Politics is not merely an exercise in estimating a population distribution. It is also a system for coordinating interests, jurisdictions, rights, obligations, and policies across differentiated communities.
the legitimate core of this is subsidiarity: allocate each decision to the constituency it affects. interior water policy that principally affects the interior belongs to an interior jurisdiction—an office defined over that population, sampled uniformly. the office defines the population; the sample matches it. that handles land use, water, extraction, infrastructure, and local jurisdiction completely, and i endorse all of it. subsidiarity never requires extra weight for interior voters inside a statewide office. you're running together decentralized jurisdiction (often welfare-positive) and disproportional weighting within a jurisdiction (the only thing i've called distorted).
I am arguing that a federal structure may reasonably represent both persons and constituent territorial communities […] to prevent persistent statewide majorities concentrated in some regions from overriding sufficient political investment in smaller constituent regions.
if persistent-majority protection is the motive, notice geography is one cleavage among many. renters, the disabled, religious minorities, workers in a dying industry—all face persistent majorities. territorial standing protects exactly one minority type, the spatially clustered one, and dilutes the rest to do it.
and see what kind of move this is in general. address is an attribute. weighting by address is structurally identical to weighting by race, by religion, by ideology, by how deontological someone is, or by fondness for spicy food if we want to be pedantic: pick a trait, and give the people who hold it more than their per-person share of representation. "these people deserve extra say" is the entire content, however it's dressed. the escape you'll want is "geography is different because it correlates with distinct interests—land, water, jurisdiction." but every demographic attribute correlates with distinct interests. race does. occupation does. religion does. if correlated-interests justified a multiplier, it would justify a racial multiplier, and you don't believe that, so you need a principled line between geography and race as weighting bases, and at the level of "correlates with interests" there isn't one. this isn't a novel observation, either—it's holding law. reynolds v. sims (1964) struck down alabama's geographically weighted state senate ("legislators represent people, not trees or acres") and explicitly rejected the federal analogy: states may not mirror the u.s. senate's territorial weighting, because counties were never sovereign parties to a compact. territorial weighting in a state-level body is the exact thing the court held unjustifiable, for the exact reason: the 1787 bargain was an instrumental deal—no union without it, and union beat disunion for everyone—and it justified the federal senate and nothing else.
Geographic representation can certainly be designed badly. Districts may be unequal, arbitrary, or gerrymandered. But none of that proves that territorial representation is inherently irrational […]
follow your own concession. require districts population-proportional and uniformly sampled within—the only version that isn't "unequal, arbitrary, or gerrymandered"—and you have arrived at my position: proportional stratification, which weights nobody and which i have zero objection to. everything beyond it is disproportion, which is the thing under dispute.
A person who did not know whether they would belong to a dense metropolitan majority or a smaller rural region might reasonably prefer a system that protects both individual equality and some degree of territorial standing.
"might reasonably prefer" is the claim restated, and it needs the one thing the veil removes: knowledge of where you live. name what the veil agent gains ex ante from the geographic multiplier. insurance? dominated by intensity-sensitivity, above. protection of the worst-off? the worst-off are at least as likely urban renters or the disabled as rural landowners, and your multiplier dilutes them. there's no candidate left.
The proper object of representation cannot be derived from sampling theory alone […] You have assumed a political objective and then labeled alternatives statistically inaccurate relative to that assumed objective.
if you believe a live instrumental case exists—that address-weighting raises total welfare in a state-level body today, where reynolds found none—state it as the empirical claim it is and we can evaluate it. it can't be derived from "coordination" in the abstract.
If you believe my position contains a contradiction, identify the propositions that generate it.
a position doesn't need a contradiction to be false; unsound premises suffice. yours requires that some persons count more than others by address, and every justification you've offered for it—insurance, minority protection, coordination—is served better by a mechanism that weights nobody. you asked which premise i reject: none. symmetry, pareto, instrumental rationality—i accept all of them, and equal weighting follows. now name yours.
-
@clay, this has become too sprawling to address productively all at once, so I’m going to isolate one point.
You write:
“Your alternative target—‘representation of constituent territorial units’—amounts to a different office, one answering to a federation of localities. That’s the boundary question again, which is upstream politics, and which I’ve never claimed the mechanism settles.”
That appears to concede the point I have been making. Whether a higher-level office represents individuals in one statewide constituency or instead coordinates constituent localities is an upstream political and constitutional choice. Sampling theory can evaluate a jury relative to whichever target is chosen, but it does not choose the target.
You may argue that the statewide individual-population target is preferable. I may argue that a more hierarchical or federated structure deserves consideration. That is a substantive institutional disagreement, not evidence that either side is statistically illiterate.
I am not going to respond simultaneously to every additional argument about score voting, Harsanyi, Reynolds v. Sims, subsidiarity, minority protection, and utility aggregation. That produces parallel monologues rather than examination. We should settle this first point before moving to the next one.