design of democracy (4)
numerator and denominator

LTNE. No comments since last time so either 1) i'm making sense or 2) no one gets it or 3) no one's reading. I'll err on the latter and push on to see where this design exercise leads me.

I left the last post with this key equation:


The numerator (v) is the number of voters. The denominator (r) is the amount of resources a candidate has invested into getting (re)elected. The quotient (c) is the acquisition cost per voter. (OK, for the math purists, we can call them dividend and divisor.)

Jay W. Forrester, used to say that people know intuitively where the leverage points are in a system -and that a lot of attention is put on pushing that leverage point, but often in the wrong direction.

Our electoral system is no exception. We know that overaccumulations of r is detrimental to our system. Our ideal is that candidates stand and are judged on the basis of their ideas or proposal. The reality is, they get elected based on name recall and on meticulously crafted media images.

We have thrown regulation after rule after law to try to corral r -trying to limit it by imposing ceilings on campaign expenses, or trying to balance it by providing government subsidies. We try to wrestle the positive feedback gain so that it doesn't become an advantage. In design terms, we have spent the better part of our energy trying to collar this factor using external controls.

Acting on r -the denominator - is pushing in the wrong direction. What we have not considered is the numerator. We have not considered the size of v nor have we considered managing v to moderate r.

Think about it: our democracy (as all democracies) were designed for much smaller populations. (Athenian democracy ran at the city level.) It did not foresee elections with millions of voters -nor did it comprehend the power of mass media.

If anything, we have let v get away without any controls whatsover. We delineate v not based on population size -but on geography. It doesn't matter whether one district has triple the number of voters of another district -they still both get one representative. The geographic boundaries are totally arbitrary. The district outlines are artificial -and have no grounding on population flows.

We ignore the effects of the size of v in our system to our detriment.

Compare the relative effect of spending P1M vs. P13M* on a campaign with a universe of 50M votes. The cost of acquisition per voter (c) for the candidate spending P1M is 50 -while it is only 4 for the candidate spending P13M. A cost difference of 46M.

But what if v=2,500? The difference in c between a candidate who spent P62,500 vs. a candidate who spent P8M is 0.0397 (corrected thanks to Willy Priles) - that is to say, almost negligible!

Our work should be on the numerator -and the size of the denominator will almost be insignificant. That is to say, work on the size of the voting population and the amount of resources one candidate will have (how much name recall they have or how big their political/social network is) will probably not matter.

NEXT: The advantages of a system built on a low v.

Image credit: Aqui-ali's flickr stream.

*i'm using peso values as stand-ins


Willy B. Prilles, Jr. said...

You are coming in perfectly clear, Urbano. In fact, I had been waiting for this post, and the next, with bated breath.

I sense where you are going to. Indeed, at the city level, which is what Athens was and Naga is to some extent, there is a higher probability that most everybody, to a closer degree of separation, would know everyone. So, there would be less need for the soundbite that most national politicians are being judged of their worthiness.

I am looking forward to how you will tie this up -- from the perspective of design -- to the form and structure of government that would best fit a country with millions of voters like ours.

Willy B. Prilles, Jr. said...

I have questions though on the example involving a 2500 universe of voters. I opened my spreadsheet, plugged the figures and got figures that don't match yours.

The difference in c between one who spends 62500 and another who spends 8M would be 0.0397, not 0.0022. Maybe I got my formulas wrong?

Urbano dela Cruz said...

thanks willy for patiently waiting for the "susunod na kabanata."

I think you're sense is right - my design investigation is pointing to a question of granularity. but more on that later.

and thanks for catching my faux pas. It was late at night and I was reading too quickly from my spreadsheet. 0.0022 is the difference between spending 1M and 8M. 0.0398 is the difference between spending 62.5K and 13M. (further down, the difference between spending 62.5K and 34M is just a hair up from 13M -- 0.0399!)

koikaze said...

Hi Urbano,

I'm almost ashamed to ask this, but the points you are making are so imporant I want to be sure I understand correctly. I'm having difficulty with the formula v/r = c. I'm sure I'm looking at it wrong, but I can't figure out why.

If I'm campaigning in a community of 2,500 voters and expend the equivalent of $1,000 worth of advertising and promotion (whether because of my job or from the actual purchase of materials), have I not incurred a cost of 40-cents per voter? That is, the amount (r) divided by the number of voters (v) gives the cost per voter.

When you divide the number of voters (v) by the amount (r) (2500 / 1000) you get 2.50, but I'm not sure what that value represents.

If the number of voters is cut in half (to 1250) because the community is split into two parts, and if I still spend $1,000, my cost per voter rises to 80-cents per voter. Calculated with v/r=c, the value drops from 2.50 to 1.25, which is your point. I think my problem is that I don't understand what dropped. It's not the cost per voter; that increased because I spent the same amount of money while appealing to fewer voters.

I've clearly got something backwards, but I don't know what it is. If I'm obtuse, please just say so and pass on.

I think you are saying the solution to the problem of candidates getting "... elected based on name recall and on meticulously crafted media images" is to increase the number of representatives for a given population size. I believe that is a valid assertion, with the caveat that its efficacy will vary with the ratio of voters to representatives.

In the U. S. House of Representataives, each member represents an average of something over 500,000 people based on population and probably something over 1,000,000 based on registered voters. Whatever the actual numbers, they are nowhere near adequate to diminish the effect of the resources candidates amass. As a result, I believe well over 80% of incumbents get re-elected ... ample support for your thesis (whether I understand the formula, or not).


p.s. I am working on several other responses to your posts and will post them as soon as I can. As you can imagine, I'm anxious to get to the questions you asked about the whatchamacallit / triad / troika method. One quickie I can't resist: You don't have to corrupt two people, only one!!! flg.

Urbano dela Cruz said...


the number of voters in the electorate for two candidates running in the same district does not change. (i.e. -theoretical district 1 will have x number of voters -that holds constant for two candidates in the same district in the same election.)

c is the cost of voter acquisition and I use it only to compare the relative effect of resources -it is a comparable ratio if you hold V equal between two candidates and allow for variances in r (resources they expend).

the equation shows that a variance in resources has a bigger effect on a large electorate. e.g. -more money, more effective campaigns. (!!! -there's that word again.)

variances in money don't make much of a difference (i.e. -smaller differences in C) when the v is small.

which is to say, throwing more money into a campaign won't help much if there are only 2,500 voters.

which effectively blunts the resource/money advantage -and levels the playing field.

Essa Remoquillo said...

OK, I must admit getting confused by LTNE. I was wondering what the reference to a low-technology nuclear explosive was doing on this page.

Gripping stuff, Benj. I realize I am commenting on a rather dated entry, but I wish I saw it sooner to assure you there are interested folks out here who hear the gong you're sounding very clearly.

Quick Links

Notable posts on the metro