This is inspired by and in response to @jcm80’s post Next steps for the GOSH Governance Working Group, namely point “3. How decisions are made and discussed.”
I am interested in how to make the decision-making process more inclusive of diverse opinions, especially how inclusiveness can be better built into polls and/or elections. Recently there has been a discussion on the Snowdrift.coop forums (a community developing ways to make open source projects financially sustainable!) on this topic in the context of their community governance, and I think it would be worthwhile to think about that here, too.
Hopefully the content and location of this post are appropriate and relevant to GoSH. Please excuse and correct me if I’m wrong, but if not then here goes:
The problem
A classic critique of polls (e.g. to choose something) or elections (e.g. choosing a person) is when participants are asked to make one - and only one - choice among several options. The winner is the option that has received the most votes (regardless of how many votes other options received). This is known as plurality or simple majority voting, which often leads to results where the winning option does not have support from most of the people who voted (this is also known as “first-past-the-post (FPTP)” or “winner takes all” in some circles). For example, the winning option might have only received 30% of the votes because that’s more than those received by any other option.
If you have voted in an election, there is a high probability you are also familiar with the spoiler effect which is a quintessential side effect of plurality voting. This is where you do not choose your truly preferred option because you don’t think it will win. The fear is that by not selecting the likely winner, you might inadvertantly let your least preferred option win. Therefore, you strategically choose the option you think will win (even if you don’t like it) to prevent your least preferred option from winning. I hope it is not controversial to say that plurality voting and the resulting spoiler effect are problematic. This model for decision-making does not allow for the expression or incorporation of diverse and nuanced opinions. In other words, it is not expressive.
Solutions
The good news is that there has been a lot of discussion on how to improve polling/voting-based decision-making, and initial attempts implementing replacement systems have been made. This is an enormous topic and for a more comprehensive treatment of the options, please refer to the Snowdrift.coop governance thread. Here I will present three examples which have open source, off-the-shelf web-based implementations that GOSH can use immediately.
-
Score Then Automatic Runoff (STAR) (code repository - AGPL 3.0 license) - Here, participants assign a score to every option and the two highest scored options move to an automatic runoff round. In the runoff, the option which received the most votes wins (official explanation here). STAR polls are very expressive and the web implementation is easy to use, allows one winner or multi-winner polls, creating custom poll URLs, private polls, and randomising the order of options presented to reduce bias.
-
Condorcet Internet Voting Service (CIVS) (code repository - MIT license) - This polling method asks participants to rank (not score) as many of the options as they want and chooses a winner that would win against all other options in pairwise comparisons. The instance linked to here is hosted by Cornell University and is used by mature open source communities such as the Debian project (e.g. see this page on their 2020 project leader elections for a detailed breakdown).
-
Helios Voting (code repository - Apache 2.0 license) - This implements approval voting where participants select any and all options that they are happy with (as opposed to having only one choice). The instance linked to here claims to take ensuring the integrity of ballots extremely seriously. Approval voting is easy to administer but is considered to be less expressive than solutions 1. and 2. UPDATE: I do not recommend this instance because it seems to only allow closed-source Google- or Facebook-based logins which are horrible for privacy and would expand global mass surveillance. To use Helios Voting, one should instead host their own instance from the source code.
My suggestion
To be clear, polls/elections are not the only way (or best way?!) to make decisions, but if one were to conduct a poll or election, the solutions presented in this post are far more inclusive.
I respectfully suggest that in the relevant governance document(s), GOSH should specifically avoid the use of plurality/simple majority voting for any decision, whether to choose people or things. The text might look like:
To make the decision-making process more inclusive, GOSH will not use plurality/simple majority voting defined as any poll where the winner is the option that has received the most votes and where each vote is only allowed one choice.
To take this a step further, the text could additionally state:
Instead, [insert polling method here] will be used whenever a poll is conducted. [Define the method here].
That said, I do not know if it is a good idea to limit GOSH to a particular polling method or leave it open.
The members of the GOSH Governance Working Group are far more experienced than me in organising communities so perhaps this post is redundant or irrelevant. If so, apologies for wasting your time!
In any case, what am I missing? Looking forward to your critiques and learning from you.
P.S. I am not necessarily suggesting GOSH use this software, but just for reference Loomio is a popular self-hostable community management platform (code repository - AGPL 3.0 license) that implements several kinds of polls including score, “dot”, and ranked-choice.