Different uses for voting need different types of voting.

Optimize fair-share funding for projects

Other Rules for
Fair Share Spending

Funding projects by Loring Allocation Rule
This page describes a series of rules, starting with simple but flawed rules, to show the need for well-developed Movable Money Votes.

A previous page explained the plurality methods often used in early programs of Participatory Budgeting. It showed some simple improvements.

Simplistic Allocation Rule, SAR

What happens if we give each project a box for contributions and give each voter his share of the FS fund to distribute and move from box to box? (This is like the STV tally analogy.) A voter gives his share to his favorite project(s). Each item's contributions are totaled. Voters then move their shares to adjust each project's total. That avoids any need for the rule to transfer excess funds. (Recall that STV automatically and impartially transfers excess weight from winners.)

Allowing partial funding of a project to accumulate for a few years lets a small group of voters fund a large project. This avoids the need to eliminate the weakest items and transfer funds until all items are either fully funded or eliminated. (Recall that STV transfers weight from losers.)

Flaws of SAR

1) It lets voters spend public funds without any discussion or agreement. Like a market economy, it meets only the weakest definition of "collective judgment".
2) It ties up some funds with little or no benefit for years.
3) This requires voters to spend hours moving their money and collecting information about which projects are winning, which are losing and what their current budgets are. Like STV boxes, this leads to stalling and deadlock.
4) It gives voters strong rewards and clear opportunities for free rides, leaving other voters to pay for popular projects. That leads to bluffing, cynicism, and under funding of those public goods.
5) It may spend public funds on private goods and so confuse the electorate's concept of public goods and weaken support for collective action.

A preliminary vote could decide which items are public goods through approval voting with a [one-third] requirement. But this is not a [one-third] mandate to proceed nor a group decision about which items are cost effective -- so a better rule is needed and should require only one ballot and tally.

An item funded by 1 or 2 council members does not have a mandate; other members are not invested in its success. Thus SAR could engender an economic climate of selfishness rather than mutual support. This is not a problem if the smallest proposal costs more than each voter's share times the number of voters required for a victory.

All of these problems are solved by Movable Money Votes.

Adapting STV Software to Tally MMV

In some cases only a small change is needed in STV tally software. A voter's STV "weight" is his share of MMV money. A candidate's STV "quota" is equivalent to a project's MMV cost, the money it must win from voters. If the proposals vary in cost, the software must be able to give each project its own cost quota. Candidate elimination drops the item which has won the lowest percentage of its cost. That might not be the proposal with the fewest votes or with the least money offered.

This software may be good enough where there are over a hundred voters, and where the cheapest projects allowed costs enough to need the weight of at least [ten] percent of the voters. Clubs and home-owner associations often fit these criteria.

A simple example may illustrate. There are 100 voters and the fund has $100,000 -- so each voter's share will be $1,000. They decide that each proposal must have at least 10 supporters to qualify it as a public good worthy of public funds. Therefore the cheapest project allowed must cost $10,000 (10 voters' shares). Projects which cost more, 15,000 or 20,000 dollars, will need money from more than 10 voters.

Let's say project K costs $10,000. Now if 11 voters give it top rank, they offer more than enough money, so the project keeps roughly 91.% of each offer and gives about 9% back to each voter. 0.91 × 11. = 10.1

Another ballot gives each voter $100 and is for items costing more than $1,000 -- again the full weight of ten voters.

But dividing the projects into arbitrary cost categories is not ideal (just as corralling voters in arbitrary districts limits their democratic choices).
A) It makes someone decide in advance how much to put in each category; the individual voter can not change that on his ballots for each category.
B) It makes cheap projects compete only with other cheap projects. One category might be much more competitive then the others.
C) It leads sponsors to play tactical games with project prices, dividing or combining projects to fit the price categories.

Advanced Electoral Rules

A cunning voter can give popular proposal K second rank and give first rank to a sure loser which will be eliminated early on. If K wins on the first count, this voter does not pay for it and can give all his money to help his second choice. But if K falls short on the first count, then this voter's phony first is eliminated and his share transfers to help fund K.

Brain Meek and C.H.E. Warren each modified STV to defeat that strategy -- to make each voter helps pay for the winners he ranked highest. Their methods are needed more for fair-share spending than for proportional representation if the typical voter helps fund several projects.

Advanced Spending Rules

Using advanced STV rules for Fair-share Spending leads to a problem if the cheapest project allowed cost less than the money shared by a "politically significant minority." Such projects might not pass even a weak test for being a public good and so might not deserve public funds.

A tally must test that each project meets two requirements: the quota of 10 supporters and its cost. If the two quotas were separate, which would determine eliminations?

[ Again, a simplistic rule can use two tallies. The first ballot asks voters to approve projects as public goods. A project that wins at least [10] approvals goes on the STV ballot and that tally allows a single voter to fund it. But that does mean [10] voters feel it is worthy of public funds or that it is cost effective -- so a better rule is needed and should require only one ballot and tally. ]

We can limit the amount a voter may "offer" a project to 10% of its cost. Then if 10 voters each offer 10%, the project meets both its quota of supporters and its cost. A quota of such offers includes both money and votes.

MMV1 copies STV: each ballot votes for its current top choice only -- even if that offer uses less than the ballot's full weight. Thus a score given to a low choice cannot take weight from a high choice, until it is eliminated. This ends any temptation to cast a "punishing vote", a low score for a close rival.

Flaws in MMV1  This creates a problem: If a voter's current top choice is cheap, then most of his share is not helping any project avoid elimination. We could say this ballot is "stuck".

Under MMV1 a rep could contribute all of her money to 1 item. Chances are no 1 item is the top priority for a full majority of her constituents. So funding only 1 project would make the rep unpopular.

MMV1 is clumsy. There are several improvements which may lead to a set of winners with greater value to the voters.

An item that is first choice on few ballots may be a second or third choice on many ballots and so get offers in later rounds after transfers. Items top rated by only a few voters are eliminated quickly. Eventually some item reaches quota. Unlike STV, MMV1 must then revive all eliminated items after a win because a voter can help fund several items.

1) Tally until all projects are funded or eliminated, then exclude the last loser at the end of the last round, and run the whole tally again. The last loser had the best chance it could ever get; all higher preferences were funded or eliminated and still it could not get the funding it needed. Some voters were stuck on this last loser while their other high choices were eliminated. Perhaps some of those others could win if the "sticky item" were excluded from later tallies.]

MMVa would let a ballot give one offer to each of its top choices, for as many as it can afford. But then a ballot would give no more help to its first choice than to its third choice if it has enough money for both. That might be OK in a large electorate, but with less than 1000 voters it might fund lower choices and miss some top priorities. Also, a groups with a near quota of ballots still can get stuck on a top choice that cannot quite reach quota. Those ballot weights go unspent throughout the tally. At the end, other voters have no money left to help the group fund its lower choices.

Unsticking and __ : The tally may let the voter or ballot help several top choices at once. This can preserve the principle that later choices must not hurt the top choice.

The relation between STV and MMV is demonstrated by setting MMV software to tally an STV rule. MMV software can tally STV. Each voter's share of weight is set at one dollar. The fund of dollars is set to equal the number of voters. Each candidate's cost or Hare quota is set at the number of voters divided by the number of seats.)

There are several ways to solve this. Both respect "the principle" that later choices must not hurt the top choice. One respects "utility values" set by the voter.

The MMV tally page shows one way to modify STV so it tests each project for both breadth of support (number of voters) and for intensity of support (money from each supporter).

Of course, a voter is more likely to win his favorite if he gives it more than one vote as shown on the ballot page.  The tally page detailed how a ballot apportions money and votes among a voter's top-rated items. To play with apportioning money, download the ballot. To learn by doing it with a group, try tabletop voting. All types of voting can be improved and that is most evident in funding rules. This site divides project funding into several web pages. This page has explained the logic and arithmetic of the Loring Allocation Rule. The next 2 pages explore optional features for LAR. Sorry to say, those pages still show the creative disorder that often occurs during research and development.

The next two pages will describe inferior, non-STV methods that approximate Fair-share Spending (FS). The Minority Funding Process (MFP) is done by a show of hands. It feels tedious to some voters and limits the numbers of proposals and voters, but it has worked. The Limited Utility rule can be calculated on a simple spreadsheet.

The Other Rules page of the next chapter has more.

NEW Free software in p_fund.htm gives a feel for scores, preferred budgets, quotas, offers, and power. Download and play with the interactive ballot. Microsoft Excel 5 or higher is required.
p_options.htm explains how MMV can keep the order of winners and eliminations from making a popular item lose; how voters may avoid buying 2 of a kind in rival proposals; and how most variations on MMV reduce to the rule explained above plus a second variable with the same effect as quota.
fundRank.htm presents tables and charts showing variations on utility curves.
z_future.htm sketches ideas for other funding rules. (Humor: The latter pages show the creative disorder that occurs during evolutionary divergence when species proliferate.)
1998 Site Introduction
1998 Site Outline


Loring Allocation Rule

Search Accurate Democracy


español Chinese