Proposal: consider using a tool like Kialo for proposal discussions / debates

[theckman]

I've noticed a pattern recently of contentious proposals being hard to follow / parse, a sentiment which has resulted in some tooling being built and some asks for things to be discussed outside of GitHub.

I wanted to create a meta-proposal for considering the use of a tool like Kialo for having the discussions and visualizing the different sides of the debate around a proposal.

The proposals around error handling have been pretty contentious, with different proposals being floated, which makes it hard to really follow what's going on. I've mocked up a very high-level overview of this discussion to show how it can show the different trees of the discussion:

• https://www.kialo.com/should-error-handling-in-go-be-changed-29766?path=29766.0

I feel like it could help us see where the larger consensus is, and to more easily spot issues that may be buried in a comment somewhere.

Has anyone else seen a similar tool that could help? What are the general thoughts around this problem, and using a tool like Kialo?

[networkimprov]

A different service probably wouldn't succeed unless you can sign in via Github.

Also note that the Go team has stated that they aren't especially moved by discussion vote tallies, presumably since they have a few years of survey data sets.

[theckman]

Sure, before we pursue asking Kialo for GitHub support I'd like to gather whether there's something we're interested in. If there's an interest, I can drive a discussion with them to see if there's an appetite to add it.

In terms of the survey, I know all too well the challenges of getting meaningful results out of a survey. While the survey results can show what people want, are we explicitly asking the inverse question around if they don't want something? For example, in the survey where we asked about handling of errors I don't remember if the survey had two options:

• Change error handling in Go
• Leave error handling in Go as-is

I wonder whether the sentiment of a discussion is a better signal than surveys with implicitly biased data, assuming we've not asked the inverse questions in these surveys.

[ianlancetaylor]

With regard to Kialo itself, rather than "a tool like Kialo", I'm reluctant to ask people to use yet another proprietary tool to participate in Go language discussions. GitHub is hard to avoid these days, but it shouldn't be a precedent for adding another proprietary tool.

[rsc]

I agree very strongly with @ianlancetaylor: we should not add yet another proprietary tool.

If anyone wants to lean on GitHub to implement something more like the conversation tree view of old news readers, that'd be great. Or if anyone can build one that presents an existing GitHub conversation that way, that's also great.

[rsc]

Another possibility would be to set up a golang-proposal mailing list and have the big conversations there. The smaller things (like #32062) should stay on GitHub. But maybe at a certain point we just say "this conversation is too big for GitHub; let's move this to the golang-proposal list". And then everyone can use whatever tools they already have for managing large email discussions.

The problem with this approach would be that there is not an obviously great web view of the discussion. On the other hand, that's a problem with GitHub too, and at least Google Groups doesn't drop the middle 90% of the messages when displaying the conversation.

[rsc]

I'd be happy to hear suggestions for better ways to manage very large discussions that don't involve proprietary tools (Kialo, Slack, ...). But definitely Kialo is not an option for the reasons given above. Closing.