Talk:Allocated Score

This is the discussion page (the "Talk:" page) for the page named "Allocated Score". Please use this page to discuss the topic described in the corresponding page in the main namespace (i.e. the "Allocated Score" page here on electowiki), or visit Help:Talk to learn more about talk pages.

Frustrating

I find the Allocated Score page hugely frustrating because it is so hard to understand. Actually I've given up trying to understand it. I've read it several times and it is just hopeless.

I get the first step: "Each round elects the candidate with the highest total score." That's easy.

But this? "After each selection, the Hare quota of ballots which scored the winner the highest is allocated to them, and as such those ballots are removed from subsequent rounds." I have NO idea what that means.

And this? "After each selection, the Hare quota of ballots which scored the winner the highest is allocated to them, and as such those ballots are removed from subsequent rounds." After several reads, I thought, OK, this means that the votes that already elected people are set to zero so they won't elect anyone else. Fine. But then, how does the Hare quota come into it?

What the page desperately needs is an example. Say, a simple fake election, with 3 to 5 candidates and 1000 voters (or scores adding up to 1000 maybe?), and going step by step through each round, showing the entire process.

Sadly, it doesn't have that. It has Python code, which I for one find totally unhelpful because it involves library functions and I don't know what they do. And anyway, one shouldn't have to read code to understand an electoral system. Not everybody can read code.

I really want to understand this system, but having found no comprehensible explanation of it anywhere, I'm prepared to give up and assume that Allocated Score is just too hopelessly complicated to use in real life. I'm sure that's not the intended result, and I'm also sure I'm not the only person to look at this page and come to this conclusion.

So can someone please come up with a step-by-step example? I'd do it myself if I possessed the required understanding, but I don't. Thank you. unsigned comment by Vunger - unknown timestamp (UTC)

Hello. Perhaps I can help you understand. "After each selection, the Hare quota of ballots which scored the winner the highest is allocated to them, and as such those ballots are removed from subsequent rounds." This means that you sort the scores given to the winner and takes the voters who scored them the highest out of later rounds.

"After each selection, the Hare quota of ballots which scored the winner the highest is allocated to them, and as such those ballots are removed from subsequent rounds." This is the same statement just rephrased. The Hare quota is the number of voters to set to zero to be taken out of the election. If there are 5 winners then it is 1/5.

Of course this page does need an example and we are working on it. Its hard to know what people would understand the best. How about this? "Winners are selected in rounds. Each round elects the candidate with the highest total score and then designates one quotas worth of votes from that candidate's strongest supporters. Subsequent rounds include all voters who are not yet fully represented." --Dr. Edmonds (talk) 19:10, 16 February 2021 (UTC)

Perhaps [this will help.] That series of flowcharts explains how I derived Apportioned Score from STV --CiaranDougherty 22:48, 26 March 2021 (UTC)

Apportioned Score Voting is not the same system as Allocated Score. They are somewhat similar and the committee that designed Allocated Score was aware of Apportioned Score when they did it. However, when they reviewed Apportioned Score they found issues. I cannot recall exactly the issue but I think it was susceptible to clones. --Dr. Edmonds (talk) 21:23, 5 April 2021 (UTC)

No, actually, it's the system I invented back in 2017, only without any of the improvements I made on it since the first draft. As such, it's not just similar, it's the same, except without the fixes to minimize Hylland Freeriding/non-monotonicity, and the ability to account for Full/Empty/Non-Discriminating Ballots. --CiaranDougherty 21:42, 4 October 2021 (UTC)

This post contains a sketch of a system but no exact details. For example, do you allocate based on the original score given or the reweighted score. This matters. The system you later defined as Apportioned Score Voting is different. This sketch of a system was also played with by Toby Pereira, Clay Shentrup, Parker Friedland and myself. We all went different directions in the end as did you. All our systems are defined clearly on electowiki. What you link to is a bare bones system and is likely what most people first come up with before they get into the details. The reason the Equal Vote committee came back to a system like this is nuanced so I will not get into it here. When we did we considered dozens of minor modifications. When we were happy with it we codified the exact definition, named it and made an electowiki page to be clear. We did not claim an inventor because it would be hard to claim any inventor for this system. This page lists Sequential Monroe as a variant, you should add your system as one too --Dr. Edmonds (talk) 20:18, 5 October 2021 (UTC)

I also find it hard to understand, for what that's worth. Trying to get ChatGPT to explain it to me is helping a little. — Psephomancy (talk) 00:04, 14 May 2023 (UTC)

Links

Does anyone have any hyperlinks to other online discussion about this method in an open forum? -- RobLa (talk) 09:05, 23 February 2021 (UTC)

There where quite a few along the way. It was in many of the analyses along with other methods. The thread which resulted in the final tweak is here. --Dr. Edmonds (talk) 21:23, 5 April 2021 (UTC)

Unfortunately those links are dead and don't seem to have made it into the archive.  :( These are the only records of the thread that I can find: https://web.archive.org/web/*/https://forum.electionscience.org/t/system-chosen-by-the-wolf-committee* — Psephomancy (talk) 00:04, 14 May 2023 (UTC)

Large party bias?

While reducing the quota from Hare to Droop can lead to large party biases in methods that are based on Plurality, I'm not sure if this holds for methods that don't have a center squeeze problem. For instance, in a Left, Center, Right situation, if the multiwinner method gracefully reduces to Condorcet as the quota is reduced, it's the consensus party that would gain more seats, not the large parties L and R. User:Closed Limelike Curves, you added a reference that replacing the Hare quota by Droop would lead to a large-party bias. Could this be a centrist bias instead, since Range doesn't have a center squeeze problem? Kristomun (talk) 17:34, 8 March 2024 (UTC)

By "largest parties" I mean those with the highest total scores (i.e. sum of points). The Droop quota is the maximally-biased admissible quota; see Pukelsheim's book. Or maybe don't, since it's a great way to get nerd-sniped into spending a month learning about one-seat rounding errors. :) —Closed Limelike Curves (talk) 20:32, 8 March 2024 (UTC)