Voting for Priorities: Possible Structures of Direct Democracy

In considering the possible mechanics of a directly democratic society we should note that different decisions are different. In particular, there is a difference between decisions that involve discrete choices and those which involve a continuum.

For example, electing a representative/selecting a delegate is a discrete choice – you cannot send half of one person and half of another. Setting aside the extent to which on important and/or simple questions, they can be sent with binding mandates, so that their identity is unimportant, for all the fiddly negotiating and compromising, the elector is making a discrete choice.

What this means is that information gets lost, in particular through “wasted votes”. 60% of people vote one way, and the votes of the dissenting 40% have no effect. If this is summed over elections done in a great number of assemblies, and then summed over the different rungs of assemblies at higher and higher levels, a lot of votes will be wasted, and so there is a lot of ‘noise’ in the result.

If our goal is the accurate transmission of the popular will as ‘signal’ from individuals up to high-level decisions, we might want to avoid this. Especially because the overall effect is likely to be a distortion in favour of the norm, in favour of established forces.

This problem is seen in the ‘first-past-the-post’ voting system used in UK elections: a government can get only 30% of the vote, but win more than 50% of the parliamentary seats, because the votes of losing and smaller parties are spread out and form small sections within each constituency which are ‘wasted’. This is a representative rather than direct system (in the sense that electors have no ongoing organisations by which to actively control and dictate to their MPs) but the problem is the same. Plus the dictatorship of the bourgeoisie and all that stuff.

Correspondingly, one attempt to address the problem is ‘proportional representation’, i.e. voting without local constituencies. This however removes any local link to particular representatives – which whatever it’s significance in a representative system, is hardly much good when the local constituency is meant to actually exist as a political force and be in charge.

Hence the further suggestion of a mixed voting system, with some constituency representatives and some ‘compensating’ members. But this seems to me to be completely the wrong way to approach things. The reason for the problem was the nature of the choice being made – it was a digital choice, between discrete options, so it involved losing information. Why ‘compensate’ with another digital choice?

I would suggest that it makes far more sense to compensate, whether in a representative democracy or a direct democracy, with a decision that is continuous. As a particular example that seems promising to me, something like a preliminary budget-setting.

What I’m envisaging is something along the following lines. There is (set up by more complex means) a scheme of different options for collective expenditure (whether that’s the direction of the whole economy, or the government’s special spheres like health and income support and the army). There is also a scheme of different sources of revenue. These could be more or less detailed.

A very simple case might be “healthcare-education-defense-social support-other” next to “progressive individual taxes-regressive individual taxes-corporation taxes-borrowing” (yeah, my knowledge of all this fiddly stuff is pretty poor, sue me).

Then, for each interval of time, each citizen ‘votes’ by 1) doing a percentage allocation of resources to the different items in the first scheme, 2) doing a percentage allocation of income drawn from the different sources in the second scheme, and 3) giving a desired relative amount of spending (this would obviously be different in a capitalist society and a socialist one, and between different sorts of both).

The body tasked with drawing up a budget (whether that’s a parliament or a supreme soviet) then has to balance their figures in such a way that each of those segments receives/gives the amount of funding that has been democratically decided upon (presumably with some acceptable margin). And the total amount of spending has to correspond to the decision (again, with a margin).

The key thing is that these numerical votes can be mathematically averaged, in such a way that the number of votes ‘wasted’ is limtied by the rounding error (i.e. if you round to the nearest hundred pounds, nearest thousand pounds, nearest penny, etc.). Each voter can feel correctly that their vote had some direct influence – some few pennies are going to this sector or that which otherwise would not be.

Are there problems with such a scheme? Well, it’s not perfect, but no arrangement is. It would certainly (maybe this is the main thing I’m eager for) totally change the character of political debate, if introduced right now. Instead of personality politics, about which candidate is ‘the right kind of guy’, political activists would have to go out and argue over whether people should raise their percentage allocation to this thing or that thing – i.e. talking about issues, and about the nuts and bolts of what’s happening, not just about personalities.

Obviously it’s open to the standard “people are too damn stupid to run their own society” objection, but then so are all democratic measures. I have great faith in the stupidity (and cupidity) of politicians.

I suggest this both as a possible structure of a future socialist society (I envisage it as a sort of ligament holding the top and bottom vertebrae of layers of federated councils/assemblies together) and also as a possibility for our own society in the here and now, not as a revolutionary change on its own but as an interesting reform. I’m interested in people’s thoughts – is there a huge problem I haven’t envisaged?

7 Responses to “Voting for Priorities: Possible Structures of Direct Democracy”

  1. Dan | thesamovar Says:

    The trouble with averaging on a continuum is that you might find that the average is something really stupid. Hypothetical example, suppose you have two areas of spending A and B, and that the service provided by A needs at least £300m spent on it, and gets much better if £500m is spent on it, but in between those two doesn’t get much better, and that spending more on it than that doesn’t make much improvement. For B, suppose it needs at least £500m spent on it, but gets much better if you spend £700m on it. Now suppose you’ve got £1bn to spend. The sensible thing to do would be to EITHER spend £300m on A and £700m on B OR spend £500m on each. If we assume that the voters are reasonable, they’ll all choose one of these two options, and if we assume there’s some diversity in values, then not all of them will choose one or the other. The end result is that the system ends up spending £400m on A and £600m on B, achieving about the same as if we spent £300m on A and £500m on B, wasting £200m of the £1bn. The point is that although money is effectively continuous, there might only be a discrete set of sensible choices. This may seem a bit contrived but I suspect situations like this would occur often.

    Another way of looking at the same problem is that averaging might produce a result that would be everyone’s least preferred option. So for example if we call the situation of spending £300m/£700m on A/B X, £400m/£600m Y and £500m/500m Z then you might have that 50% of people have preferences X>Z>Y and 50% have preference Z>X>Y. Everyone hates option Y but it gets picked anyway. Paradoxes like this happen in every voting system (because of Condorcet cycles or more generally as a consequence of Arrow’s possibility theorem) but it seems to me that it would happen all the time if you base it on averaging.

    Ultimately I guess the point is, why would you expect averaging to produce good results? I mean, what is it about the average that makes it the basis of a meaningful basis for compromise? For example, I might be more willing to compromise on education spending than health spending, but if we just take averages then the compromise makes all of these the same and gives me no way to express my preference about the nature of the compromise. Do you know this quote from William Godwin (the first anarchist, perhaps)?

    The whole is then wound up, with that flagrant insult upon all reason and justice, the deciding upon truth by the casting up of numbers.

    William Godwin “Political Justice”

    There’s also the dynamics that such a scheme would introduce. If you know that your numbers are going to be averaged, do you submit what you really think would be the best way of allocating the money, or do you just jam everything in to the one thing you care about most (because that will maximise the impact of your vote on the average)? If everyone was doing that, the results could be catastrophic!

    Have you read about participatory budgeting? Google the participatory budget of Porto Allegre for an interesting example.

    Another possibility that I think is interesting and may be possible now or soon thanks to improvements in communication technology is iterated negotiation. That is, everyone starts off by submitting their numbers, they’re averaged and those numbers are transmitted back to everyone. Now you can change your vote, knowing how other people are voting. Submit your results, averages computed, transmitted back. Repeat. The idea would be that after a few rounds a consensus compromise would be reached. To force that, you’d probably have to make each round limit the possibilities of the next round. So that, for example, in round 1 any allocation is allowed. In round 2, you have to choose an allocation that is not too far from the round 1 allocation, and in round 3 you choose from options that aren’t too far from round 2, etc. You’d have to choose the number of rounds carefully, but maybe something like this lets people express their preferences about what they want to compromise on a bit more. You could also have expert commentary on each plan that people could read and base their decisions on. It wouldn’t be binding, but it would make wasted money like in my example earlier less likely because most people would adapt their preferences.

    Mike Albert’s and Robin Hahnel’s stuff on participatory economics is quite interesting on these sorts of issues.

  2. Alderson Warm-Fork Says:

    You’re right that if utility by resources graphs have funny shapes, the average could be inefficient. I think that can be largely addressed though. Iterated negotiation is one good method (thanks for that suggestion!). Another would just be to have public discussion prior to the vote itself, possibly with a prominent role for something like political parties – some number of ‘model proposals’ that can expect a substantial amount of votes, drawn up with expert input (well, presumably we get various sorts of experts talking loudly and endlessly about the whole thing) that can then get argued over. The parties wouldn’t be much like now because there’s no way you can ‘whip’ individual voters the way you ‘whip’ legislators.

    These are in a sense the same solution, to the basic problem that a rational decision is hard to make if you only have one opportunity to say anything. Just as an individual needs a chance to reflect and revise and perfect, so groups should have opportunity to revise, reflect, ‘iterate’, etc. Tricky part is when doing so takes days. But still, it seems like a bearable drawback, since as you say all systems have problems.

    I’ve read a little around participatory budgetting, and it seems like a step in the right direction, but I get the impression it still leaves ultimate decision-making in the hands of the small group of budgeters, whereas what appealed to me about this mechanism was that it puts “hard limits” on any body that’s allocating resources.

    “why would you expect averaging to produce good results?”
    Principally as a safeguard, I suppose – it’s not so much averaging per se or any other specific algorithm of aggregation, as the idea of just taking some high-level decisions out of central co-ordinator hands. If there’s only a relatively small number of categories, after all, then there’s plenty of leeway left for responding flexibly to the kind of wiggly graphs you describe. But it limits that flexibility so that if, on average, people want 5% of their budget spent on the military, and 30% on health, they don’t get 30% spent on the military, and 5% spent on health. Or to use an example presupposing socialism, if people on average want to use 60% of social capital on consumption, and 40% on investment, they don’t get them the other way around.

    I’ve come across PE, it seems like some good stuff, although I’ve not studied it in too much detail. I would say that I don’t think any single model should be imagined to be applied to all places in all times – developments would require changes. But I don’t think PE has to deny that, and it seems like a good medium-term image (where ‘long-term’ means century or so post-revolution or whatever).

  3. Dan | thesamovar Says:

    I absolutely agree that decentralising decision making seems like a good idea because clearly spending doesn’t reflect people’s priorities, and so it’s worth thinking about mechanisms that make this sort of thing possible. It’s actually one of the things I’d love to spend more time thinking about. It’s something that could make a real difference.

    But there are a lot of pitfalls. So, my example with A and B wouldn’t be helped by having expert discussion beforehand because I already assumed that everyone perfectly knows the utility graphs and that there are only two sensible positions, but that there is diversity of values in the population. The end result is still inefficient.

    I agree that if you make just one decision it’s difficult to imagine that you could get across even a tiny proportion of the complexity of your views (although maybe rank ordering alternatives might get close), and so it looks like some sort of iterated scheme would have to be involved. If the spirit of politics is about negotiation and compromise, this ought to be expected. One could say that it’s only the information transmission problem that made it impossible to have a system like this in the past (the logistical issues without something like the internet would completely rule it out).

    Steve Shalom’s idea of “participatory polity” (parpol, in analogy to parecon) is also interesting. It’s in some ways a sort of hybrid participatory-representative scheme. The idea is that everyone can vote on every issue, but that you limit the cognitive and logistical difficulties involved with that by giving people the option to give someone else your vote to cast by proxy, who can then choose to give another person their and your votes by proxy, etc. The idea is that you might think that say, Noam Chomsky is right about everything, so you could give your proxy vote on all issues to him. Or you could give it to him just for say, foreign policy issues, and to someone else for domestic issues. You could imagine that such a system could be a more representative representative system because you would essentially be giving your vote to people you trust rather than people who have a chance of getting political power. It could also be participatory because you might give your proxy to someone you know personally, and then discuss ideas with them, and they’d give it to someone they knew personally, and so on up the chain. Well, it’s also an interesting idea anyway.

  4. Alderson Warm-Fork Says:

    “my example with A and B wouldn’t be helped by having expert discussion beforehand”
    I guess in this sort of situation what’s called for is for people to take the initiative – e.g. seek some way to debate the two alternatives beforehand and then all (or a good majority) vote for that one. I think there’s no reason in principle why organised people couldn’t do this if the issue was important enough. I mean, what you’re basically pointing out is that some decisions are digital – which is of course true. But the post starts from the feeling that a lot of decisions are continuous, and normal voting methods make most sense with digital decisions.

    “Steve Shalom’s idea of “participatory polity””
    Ok so I just looked this up on Znet and found it to be awesome. That’s basically what I would sign up for and have argued for on this blog. The biggest difference is that we seem to put different emphases on how often councils should issue mandates to their delegates – but obviously sometimes they do, sometimes they don’t, so whatever. Plus he’s less keen on the right of secession, I think. But yeah, it’s nice to know that what made sense to me makes sense to others (and, I think, is borne out by a lot of actual experience).

    I didn’t come across the vote-trading bit though – I mainly just found the ‘nested councils’ stuff. Do you have a link?

  5. Dan | thesamovar Says:

    “I guess in this sort of situation what’s called for is for people to take the initiative – e.g. seek some way to debate the two alternatives beforehand and then all (or a good majority) vote for that one.”

    But that’s just begging the question though – it’s saying the voting is irrelevant because we could find some other way of reaching agreement. But what is the other way?

    “I mean, what you’re basically pointing out is that some decisions are digital”

    It’s not just that, though you’re right that that’s all my example shows. As you said, if the utility graphs are even slightly odd shaped then averaging may very well find you in an inefficient position.

    “Ok so I just looked this up on Znet and found it to be awesome.”

    Yeah, it’s been a while since I read about it but I remember thinking that it seemed pretty sensible. I just checked, and I couldn’t find anything about the vote-trading stuff either. Maybe my memory is at fault, it was a long time ago when I read about it (when it first came out). Possibly it was something that was debated on the Z sustainer forums either before or after he published his main essay about it, and maybe that idea was an early version of his final one, or possibly just someone else’s idea when discussing it. It’s quite similar to the nested councils, a bit more flexible but maybe a lot more organisationally complicated.

    Those ZNet guys have a lot of interesting ideas (that are pretty relevant to anarchism I think).

  6. Alderson Warm-Fork Says:

    “it’s saying the voting is irrelevant because we could find some other way of reaching agreement. But what is the other way?”

    Well in this case isn’t the other way a digital vote – a set of alternatives are identified, with anything intermediate between them substantially less efficient, and people vote for one of them, and the most popular is implemented. I.e. a referendum I guess. So if there are some decisions that are best made in that way, and others best made in a continuous way (as, it seems to me, rough-grained resource allocations are liable to be, on the whole) then it makes most sense to use the different voting systems in the two cases. How to decide how to decide, of course, is a further question, that will have to be approached through other methods.

  7. Dave Says:

    The key point here is not found in the specifics of your proposal, but rather in your assertion that this type of system would ellicit further discussion. The notion that the choices would need to be continuous is a bit distorted, in that what would essentially be a continuous result could be derived from a series of digital responses.

    The issue most often presented is a digital decision to be made regarding an amalgamated whole. Your budget example could be approached in a series responses designed to determine relative prioritization.

    Again, in the end, I think this is just another example that some form of direct democracy could work if given appropriate cycles to devise a methodology — thanks.

Leave a Reply

Fill in your details below or click an icon to log in: Logo

You are commenting using your account. Log Out /  Change )

Google+ photo

You are commenting using your Google+ account. Log Out /  Change )

Twitter picture

You are commenting using your Twitter account. Log Out /  Change )

Facebook photo

You are commenting using your Facebook account. Log Out /  Change )


Connecting to %s

%d bloggers like this: