12 KiB
There's no point being coy or saying anything but what I actually believe, and saying it bluntly. So here it is:
The haskell.org committee has consistently engaged in tactics which silence the voices of all non-members, and stacks the committee to prevent dissenting opinions from joining.
I've said various parts of this previously. You may have heard me say things like the haskell.org oligarchy, refer to the "evil cabal of Haskell" (referring to the nepotism which exists amongst Hackage, cabal-install, haskell.org, and the Haskell Platform), or engage in lengthy debates with committee members about their actions.
This is a pretty long post, if you want to see my request, please jump to the end.
The backstory
To summarize a quick backstory: many of us in the community have been dissatisfied with the four members of the "evil cabal" for years, and have made efforts to improve them, only to be met with opposition. One by one, some of us have been replacing these components with alternatives. Hackage's downtime led to an FP Complete mirror and more reliable doc hosting on stackage.org. cabal-install's weaknesses led to the creation of the Stack build tool. Haskell Platform's poor curation process and broken installer led to Stackage Nightly and LTS Haskell, as well some of the Stack featureset. And most recently, the haskell.org committee's poor decisions (as I'll demonstrate shortly) for website content led to resurrecting haskell-lang.org, a website devoted to actually making Haskell a more approachable language.
As you can see, at this point all four members of the evil cabal have been replaced with better options, and community discussions and user statistics indicate that most users are switching over. (For an example of statistics, have a look at the package download count on Hackage, indicating that the vast majority of users are no longer downloading packages via cabal-install+Hackage.) I frankly have no problem at all with the continued existence and usage of these four projects; if people want to spend their time on them and use what I consider to be inferior tools, let them. The only remaining pain point is that new, unsuspecting users will arrive at haskell.org download page instead of the much more intuitive haskell-lang.org get started page.
EDIT Ignore that bit about the download statistics, it's apparently due to the CDN usage on Hackage. Instead, one need only look at how often a question about Haskell Platform is answered with "don't do that, use Stack instead." For a great example, see the discussion of the Rust Platform.
The newest attempt
Alright, with that out of the way, why am I writing this blog post now? It's due to this post on the Haskell-community mailing list, proposing promoting the Haskell Platform above all other options (yet again). Never heard of that mailing list? That's not particularly surprising. That mailing list was created in response to a series of complaints by me, claiming that the haskell.org committee acted in a secretive way and ignored all community input. The response to this was, instead of listening to the many community discussions already occuring on Twitter and Reddit, to create a brand new mailing list, have an echo chamber of people sympathetic to Evil Cabal thought, and insist that "real" community discussions go on there.
We're seeing this process work exactly as the committee wants. Let me demonstrate clearly how. At the time of writing this blog post, three people have voted in favor of promoting the HP on haskell-community, including two haskell.org committee members (Adam Foltzer and John Wiegley) and the person who originally proposed it, Jason Dagit. There were two objections: Chris Allen and myself. So with a sample size of 5, we see that 60% of the community wants the HP.
The lie
A few hours after this mailing list post, I put out a poll on Twitter. At time of writing (4 hours or so into the poll), we have 122 votes, with 85% in favor of Stack, and 15% in favor of some flavor of the Haskell Platform (or, as we'll now be calling, the Perfect Haskell Platform). Before anyone gets too excited: yes, a poll of my Twitter followers is obviously a biased sample, but no more biased than the haskell-community list. My real point is this:
The haskell.org committee is posing questions of significant importance in echo chambers where they'll get the response they want from a small group of people, instead of engaging the community correctly on platforms that make participation easy.
This isn't the first time this has happened. When we last discussed haskell.org download page content, a similar phenonmonon occurred. Magically, the haskell-community discussion had a bias in favor of the Haskell Platform. In response, I created a Google Form, and Stack was the clear victor:
Yet despite this clear feedback, the committee went ahead with putting minimal installers at the top, not Stack (they weren't quite brazen enough to put the Perfect Haskell Platform at the top or even above Stack, for which I am grateful).
Proper behavior
As I see it, the haskell.org committee has two correct options to move forward with making the download page decision:
- Accept the votes from my Twitter poll in addition to the haskell-community votes
- Decide that my poll is invalid for some reason, and do a proper poll of the community, with proper advertisement on Reddit, Twitter, the more popular mailing lists, etc
If past behavior is any indication though, I predict a third outcome: stating that the only valid form of feedback is on the haskell-community mailing list, ignore the clear community groundswell against their decisions, and continue to make unilateral, oligarchic decisions. Namely: promote the Haskell Platform, thereby misleading all unfortunate new Haskellers who end up at haskell.org instead the much better haskell-lang.org.
Further evidence
Everyone's always asking me for more of the details on what's gone on here, especially given how some people vilify my actions. I've never felt comfortable putting that kind of content on blogs shared with other authors when some of those others don't want me to call out the negative actions. However, thankfully I now have my own blog to state this from. This won't include every punch thrown in this long and sordid saga, but hopefully will give a much better idea of what's going on here.
-
Not only are conversations held in private by the committee, but:
- Their private nature is used to shut down commentary on committee actions
- There is open deception about what was actually discussed in private
Evidence: see this troubling Reddit thread. I made the (very true) claim that Gershom made a unilateral decision about the downloads page. You can see the evidence of this where he made that decision. Adam Foltzer tried to call my claim false, and ultimately Gershom himself confirmed I was correct. Adam then claimed offense at this whole discussion and backed out.
-
When I proposed making Stack the preferred download option (at a time when Stack did not appear at all on haskell.org), Gershom summarilly closed the pull request. I have referenced this pull request many times. I don't believe any well intentioned person can read that long discussion and believe that the haskell.org committee has a healthy process for maintaining a community website.
-
At no point in any of these discussions has the committee opened up discussion to either the technical advantages of the HP vs Stack, or the relative popularity. Instead, we get discussions of committee process, internal votes, an inability to make changes at certain periods of time based on previously made and undocumented decisions.
-
We often hear statements from committee members about the strong support for their actions, or lack of controversy on an issue. These claims are many times patently false to any objective third party. For example, Gershom claimed that the pull request #122 that he unilaterally decided to merge was "thought to be entirely mundane and uncontroversial." Everyone is welcome to read the Reddit discussion and decide if Gershom is giving a fair summary or not.
-
Chris Done - a coworker of mine - spent his own time on creating the first haskell-lang.org, due to his unhappiness with the homepage at that time. His new site was met with much enthusiasm, and he was pressured by many to get it onto haskell.org itself. What ensued was almost a year of pain working out the details, having content changed to match the evil cabal narrative, and eventually a rollout. At the end of this, Chris was - without any given reason - not admitted to the haskell.org committee, denying him access to share an opinion on what should be on the site he designed and created.
My request
Thank you for either getting through all of that, or skipping to this final section. Here's my request: so many people have told me that they feel disenfranchised by these false-flag "community" processes, and just give up on speaking up. This allows the negative behavior we've seen dominate the evil cabal in Haskell for so long. If you've already moved on to Stack and Stackage yourself, you're mostly free of this cabal. I'm asking you to think of the next generation of Haskell users, and speak up.
Most powerful course of action: subscribe to the haskell-community mailing list and speak out about how the committee has handled the downloads page. Don't just echo my message here: say what you believe. If you think they've done a good job, then say so. If you think (like I do) that they've done a bad job, and are misleading users with their decisions, say that.
Next best: comment about this on Reddit or Twitter. Get your voice out there and be heard, even if it isn't in the haskell.org committee echo chamber.
In addition to that: expect me to put out more polls on Twitter and possibly elsewhere. Please vote! We've let a select few make damaging decisions for too long, make your voice heard. I'm confident that we will have a more user-friendly Haskell experience if we actually start listening to users.
And finally: as long as it is being mismanaged, steer people away from haskell.org. This is why we created haskell-lang.org. Link to it, tell your friends about it, warn people away from haskell.org, and maybe even help improve its content.
Archive link of the Reddit and Github threads quoted above: