Follow the reluctant adventures in the life of a Welsh astrophysicist sent around the world for some reason, wherein I photograph potatoes and destroy galaxies in the name of science. And don't forget about my website,

Thursday 4 May 2017

The Political Drake Equation

How do you decide who to vote for ? I've no idea, because I don't know who you are. So I shall tell you how I decide who to vote for, and then you can decide if this is sensible or if you have a better system.

Astronomy has this famous thing called the Drake Equation, which is way of estimating how many intelligent alien civilisations might be around for us to talk to. As equations go, it's tremendously simple - nothing more than multiplication. It looks like this :

It's literally just multiplying a bunch of numbers together  - say there are a million planets in our Galaxy but only a tenth are like Earth and only a tenth of those actually have life, then that's 10,000 planets with life. Easy peasy. Of course, working out what those numbers are is much, much more difficult.

A modified version of the Drake Equation makes for a pretty good way of explaining how I decide who to vote for. I don't normally actually set about plugging in numbers on a calculator, but this pretty well approximates (I think) what I'm doing unconsciously... and I suspect this is true in general as well. It might also be a handy way of explaining to people how you made your decision in (if we're very lucky) a less chest-thumping way. Maybe it'll even help you analyse your own thinking...

Anyway, the Political Drake Equation looks something like this :

V = I * P * B* A * E * R * S

V is a number which determines how much credibility you should give to the prospect of voting for any given party. The higher the number the better, but it all depends on how each party fares - you have to evaluate this equation for all parties and consider their relative scores. Only if they all get an equal voting score should you consider not voting at all (let's ignore tactical voting considerations for the moment).

Each of the other parameters represents some reason you have to vote for or against that party - low numbers mean you shouldn't vote for them, high numbers mean you should. To keep things simple let's let those numbers all run from 0 to 10. This makes the minimum overall V score zero and the maximum ten million. But a score in the millions would be very rare, and the score unfortunately doesn't vary in a nice proportional way - but absolute values are far less important than the ranks here.

Still, to get a feel for the numbers, if you assigned equal values to each parameter, V would just be x7 where x is the value of each parameter. You can see an interactive plot of this here (or make do with the static one below, if you want). A totally useless party with all parameters of 2.5 would get a score of about 600, a mediocre one with all parameters equal to 5 gets about 80,000, a really good one with 7.5 in each category gets 1.3 million. The values differ dramatically, but, to emphasise this point, even if all parties score badly this doesn't mean you can't pick the lesser of evils unless their scores are all very similar.

Because everything is multiplied, a value of zero for any reason means you definitely shouldn't vote for that party regardless of any other considerations. This is because I consider all these parameters to be absolutely essential, but in practise, it should be very rare that you ever actually give a party of score of 0 or 10 for anything (but you can give very extreme fractional values, of course, like 0.001 or 9.999). Of course, it's open to debate if each category should really allow the same maximum value, but this'll do as a start.

What I like about this is that it accounts for much more than just stated policies. Policies are irrelevant if you don't trust the party or if you think they don't have the ability to enact them. Of course, the downside is that this is necessarily a self-analysis : no-one can objectively measure how much you trust a party. But self-analyses are extremely useful as long as you make a sincere effort to understand why you really came to a conclusion, and this equation has the additional advantage of being flexible and easily modified.

The criteria I've chosen for it are as follows. When evaluating them, keep in mind precisely who each party would likely elect to power. For example, don't give a party a high Ability score just because you think they have lots of talented people - rate this value according to who you think they're actually likely to put in charge. And remember, you're not trying to rate each party by some absolute system, but only in terms of how closely their ideas and values match your own - or even just by how well the match the characteristics of a party you want in government (for example you might be a personal pacifist, but accept that political leaders generally have to act less perfectly). Anyway, here are the assessment criteria :

I = Idealism
Do the party's ideals match your own ? This is more fundamental than current stated policies, it's more about the soul of the party - if it has one. Do you believe they are, overall, trying to do things you basically agree with, even if you don't accept some individual policies ? An extreme example might be agreeing with the policies of an avowedly religious party while not being part of that religion yourself. You could also think of this as the long-term "climate" of a party, as opposed to its short-term "weather" of its specific policies.
You'll want to weight this one appropriately. For example, if you don't agree with their ideology but only care about specific policy, increase this score to compensate.

P = Policies
Do you agree with the party's policies - are they practical, sensible ways of implementing their ideals ? Unlike many of the other parameters you can objectively measure how much you agree with their policies, including the weighting for the importance of each policy (this is very important !), through simple testing. This one (click the country at the top right) gives you a score as a percentage, so just divide it by ten to use it in this equation. This should also give you at least a handle on ideology, though ultimately only you can decide if you truly agree with the party's ideals or not.

A = Ability
Does this party have the necessary skills in order to implement their stated policies ? If you don't think they're capable of fulfilling their grandiose promises then there's no point voting for them.

T = Trust
You may think that your party has the right ideology, its policies are correct and its members are highly intelligent... but this doesn't mean you trust them. Maybe they could do what they say, but you don't think they actually will. Especially for smaller parties, this can be more complex than whether you think party members are decent people - you might think that when they got into government they might be forced to make compromises they didn't want to make.

B = Behaviour
Does the party behave in a sensible way in other aspects not related to its stated principles and governmental policies ? For example, does it select its representatives for government in a way you agree with ? Does it allow members to vote freely or does it enforce votes based on party policy, and do you agree with this ? Does it manage itself well ? Do they try and convince undecided people using well-reasoned arguments or by appeals to base emotion ? If they adopt a, "the beatings will continue until morale improves" approach, then for me there'd be no point voting for such a party.

E = Evidence
Will this party act appropriately as new evidence is presented ? That is, will it change policy if the evidence goes against it or devise new policies accordingly, and will it do so because of the evidence rather than to satisfy voters ? Another term might be "sincerity". Because, you see, the evidence on at least some policies most certainly will change, so I consider it essential that a party has the flexibility to be able to deal with that.

R = Respect
Will this party treat its opponents with the appropriate degree of respect ? Will it seek to gain unfair advantage to crush its rivals or will it try and build a genuine consensus - insofar as that's possible - through persuasion, negotiation and compromise ? In your judgement, does it act correctly to deal with those who continue to disagree with party policy after all attempts at persuasion have been exhausted ? Of course, this doesn't necessarily preclude being extremely harsh to its detractors - as with all the parameters, the question is whether you think this is a good thing or not.

A Worked Example

So, let's apply this to the real world. Here are my own numbers for the major political parties in the UK. I used numbers from isidewith for the Policies index. The rest are of course my own judgement - part of the usefulness here is to be able to communicate reasons as well as self-analyse.
EDIT : I considered a way to account for the importance users attach to each parameter in the discussion here. I'm not entirely happy with the result - it still needs a bit of work, though it's probably close to what I was aiming at. Weighting each parameter makes things more complicated since it makes the maximum possible value unavoidably variable. A better approach for the end result, which probably needs to be used anyone, is to make the final score a percentage of the maximum possible, or better yet a percentage of the maximums score actually achieved.
Should I ever discover a way to create interactive spreadsheets for multiple users, I'll try and solve this and update this post accordingly.

Now of course you have only my word for it, but while I've already decided to vote for the Liberal Democrats, I was surprised at the incredibly decisive result in their favour. Especially so since Labour do better on policy according to this. I was expecting to have to artificially reduce the policy scores of both the Tories and Labour because I don't think the importance of Brexit is adequately utilised in the isidewith measurements; I'd probably opt to roughly halve the policy scores of every party apart from the Lib Dems. In my view, trying to get a good deal on Brexit won't work and so all other parties are advocating for madness. This more-or-less wipes out all of their other policies since we'd be spending years up the proverbial creek - it doesn't really matter if you've got a paddle or not if your boat is leaking and the piranhas are circling.

What you can also see here is that some parties do badly (in my estimation) on a range of factors, while some have just a few critical weaknesses. UKIP are irredeemable - I hate pretty nearly everything about them. Plaid Cymru do well in many areas but fail largely on evidence and ability : in my opinion, Welsh independence/nationalism is utterly barking bad, and they haven't convinced me they know anything much about their other policies either. A subtle point that the numbers can't show is that (especially for the Greens) sometimes I don't necessarily disagree with their conclusions, I just think that they're far more ideologically driven than evidence-based. Had the Greens persuaded me that their policies really were driven by the evidence, they'd be competitive with the Lib Dems.

Labour fail for me this time round based largely on their "leader". He has destroyed my trust that the party will do what they say, never mind whether their policies are a good thing. I don't think he's an intelligent man - he doesn't seem to have changed his views in his life - and I think he deals incredibly poorly with people who disagree with him. For me, the "I'm not leaving" after losing the vote of no confidence was a point of no return : I really just do not - indeed, cannot - understand how anyone can think such a man is trustworthy after that. But then, one man's democratic vote is another man's coup... Anyway, that's why Labour do badly in so many parameters.

The Liberal Democrats do surprisingly well here because they have no major weak points. This raises the obvious question : have I over-estimated any of these points, giving them an unfairly high score ? Or indeed have I under-estimated any points from the other parties ?

Of course there's margin of error on all of these. I won't go through them all because that'd take all day, but a few points are worth mentioning :
  • I'd consider reducing the P value for all parties except the Liberal Democrats and UKIP based on their stated Brexit stance and the exceptionally high importance I place on this.
  • I could perhaps increase my ratings for the Tories ability and trust scores with a new leader and cabinet, but I doubt I'd every change any other of their scores much so they'd always get a low overall result. 
  • Labour could improve all of their weak points with a better leader and shadow cabinet - they have by far the most to gain here.
  • The two points of the Lib Dems I'm most flexible on are ability and trust - I could knock them down to 5 and 4 respectively, but that still gives them a whopping 367,696. 
  • The Greens have very plausible scope for improvement - they need leaders more skilled in rhetoric to convince me of their abilities and critical thinking skills.
  • The SNP and Plaid Cymru are unsalvageable because I believe independence is in both cases a mad idea, just madder for Wales because we're a silly place.
  • UKIP are the sort of "party" at which everyone gets drunk and goes home with loss of limb and several terminal sexually transmitted diseases, or in other words oh God no never.
In short, this time round there's no point me considering anyone besides the Liberal Democrats, because all the other parties fall far short. Next time, Labour could plausibly be re-aligned with my own views, as could the Greens. Whether that will actually happen or if the parties all just collapse entirely remains to be seen.


What I haven't considered here is tactical voting : should you vote for the party you agree with based on the party as a whole - the combination of policy, idealism, and pragmatism described above - or on the basis of which one you think has more of a chance of getting elected ?

That's much harder. The simplest way to modify the equation would be to add a T parameter which describes how likely it is you think your vote will actually help them get elected. Then if you find that this reduces V for a party to a level below that of others you also find acceptable, you should consider voting for one of those instead. But I dislike this kind of logic - it promotes groupthink, so that the smaller parties get much less of a chance for breakthroughs.

It's also not easy to factor in the the nationwide voting probabilities versus vote in a particular constituency. For instance, say you hate the Green party and you're in a Labour-Green marginal. You really want to vote Tory but they only get 2% of the vote in that constituency. Labour, the polls indicate, have a real shot at winning the whole election but you hate them too, though not as much as the Greens. Should you vote for Labour (to prevent the loathed Greens winning) or the Greens (to help deny Labour a government) or stick with your principles and vote Tory ? Not an easy decision.

One way to proceed might be to construct three list : the parties as ranked by the PDE, the opinion polls for your constituency, and the nationwide opinion polls. Then you have to choose the party which if it wins will a) produce a nationwide result you'd find acceptable, or least awful; b) has the greatest chance of winning in your constituency, or is least unlikely from the parties you'd consider voting for and c) is the "best of the rest" according to the PDE. Satisfying all three criteria gets tricky, and is highly vulnerable to errors in local opinion polls. Still, in principle the PDE can also be used to help you choose the least bad option, though personally if I think every party is genuinely and similarly shite I'd advocate not voting for any of them.

To my mind, tactical voting often makes good sense, but not always. If the political situation is relatively stable - no unusually large issues dominating the scene, only small changes in Parliament predicted - then it probably does make sense to vote tactically - either to deny the government you don't want a seat, or to elect the lesser of two evils. But if it isn't stable, if there's a pivotal issue in the air, then personally I think tactical voting is less sensible. Only one party is saying what I support on the most important political issue right now - none of the others come close. The political scene, I believe, is not at all stable. And if I were to adjust the P index to account more accurately for the importance of Brexit, I believe the Labour and Tory votes would be far closer to each other. So voting for Labour barely gets me a party that's any better than the Tories, overall - even though the majority of Labour policies are far superior to Tory ones, in my view.


Good God no ! You must decide for yourself if any of this has been useful or merely a way to justify my decision. That's the risk of self-analyses, of course : if you do them right you can realise that you've come to a stupid conclusion and change your mind, but if you do them wrong you just make your own filter bubble stronger.

If you want to try this exercise for yourself, you shouldn't be at all worried if you get a result which disagrees with what you expect. If that happens, what you should try to do is to consider each parameter again and try and decide if you've rated it correctly. It's absolutely fine to tweak parameters to get the result you want... provided you think carefully about why you're doing so. The main goal is to get you to carefully consider how you've formed your own conclusions. And of course, if you do think you've set the numbers correctly, maybe it's time to re-evaluate who you'll be voting for.

A caveat that I've only mentioned in passing is the relative importance of each parameter. You might think that policies are absolutely paramount and the others count for nothing, but other people might think that all the criteria have roughly equal value. Currently the PDE doesn't account for this, but I'm working on a way to implement it. You should be able to effectively remove certain parameters if you want - you'll be able to rank parties just by ability and policies if you so desire. This is in development, so watch this space.

While of course the qualities I've suggested reflect my own views on how to vote, weighting would be able to compensate for that (do please suggest any other fundamental properties you think I've missed). For example, which part of "representative democracy" do you prefer ?
Those who favour democracy believe in direct rule by the people, with politicians as empty vessels that serve only to enact the will of the people. These people would rate Policies very highly while giving very low or zero scores to Idealism and Evidence, as these two qualities are decided by the people, not the politicians. They would probably rate Respect, Evidence and Behaviour rather poorly as well - they want politicians to do as they're told, not think for themselves.
Conversely, those who favour representation want the politicians to speak for their values rather than specific policies, so they would weight Policies very low while giving Idealism the highest rating, and probably giving Respect, Evidence and Behaviour high scores. They want politicians who are more expert than themselves at dealing with the specifics but who fundamentally have their best interests and ideologies at heart. Most people, of course, are probably somewhere between the two extremes.

As to my own numbers you, dear reader, have basically three possible responses. The first is to examine the method and suggest improvements or dismiss it entirely. If you find that it's basically sound, your second option (if you disagree with my voting choice) is to persuade me why I'm wrong about the individual criteria for the other parties. If you agree with both the method and (roughly) the numerical values I've assigned, then your only remaining option to change my voting preference is to explain the flaw in my tactic voting argument. So at the very least, what this does is show as clearly as possible how I've reached my decision and lays out precisely what you can do to change my mind. Good luck.

No comments:

Post a Comment

Due to a small but consistent influx of spam, comments will now be checked before publishing. Only egregious spam/illegal/racist crap will be disapproved, everything else will be published.