Dining philosophers rephrasing

[lambda]

The dining philosophers problem setup is quoted from C. A. R. Hoare's
"Communicating Sequential Processes." This is not the paper in which
the problem originated, nor does the example use the solution from
Hoare's paper. Furthermore, the quoted example uses forks and pasta,
and it doesn't make much sense to need two forks to eat; the formulation
of this problem that I am most familiar with uses chopsticks, which
makes much more sense.

In addition, the quoted example uses male pronouns, the rest of the
article uses singular "they", and refers to the philosophers with just
numbers at first, but then goes on to use five Western European male
philosophers.

Since the particular quotation is not essential, the use of forks is not
particularly sensible, and to increase the diversity of philosophers,
simply rewrite the statement of the problem with original text. Use the
appropriate pronouns for the philosophers referred to, but continue to
use singular "they" when talking about an unspecified philosopher.

[rust-highfive]

r? @brson

(rust_highfive has picked a reviewer for you, use r? to override)

[lambda]

r? @steveklabnik

[bors]

☔ The latest upstream changes (presumably #25624) made this pull request unmergeable. Please resolve the merge conflicts.

[lambda]

By the way, I agree with the reasoning behind #25585, I just thought that it made the example even more disjointed from the rest of the chapter, using feminine pronouns in the quote, then a set of philosophers who were all male, except for the one time they were referred to by number, while being referred to with singular "they." I thought that this would help both resolve that issue, and clean up the example in general.

[Manishearth]

r=me, but I'd like Steve to take a look at it too since he's been handling this and might have further commentary. Also there were substantiated comments in the original PR about preferring she over they, which get un-addressed with this; not too fond of that bit. (This is still a good compromise though!)

+1 for coming up with this 😄

[lambda]

Yes, I'm fine with switching to using consistently feminine names an pronouns as well (whether using real philosophers, or fictional), for the reasons described in #25585. This proposal is just one possibility; I mostly proposed it this way to remove the inconsistency and awkwardness of the edited quote (especially given that the actual list of philosophers below didn't match the change to the quote), rather than a particular attachment to a set of philosophers or pronouns to choose. If anyone has a preferred set of philosophers, let me know (but please not too much bike shedding).

[Manishearth]

Agreed on the bikeshedding. I personally feel that we shouldn't be overthinking this; if a doc PR is going to help a particular group we shouldn't really dither too much on technicalities (unless it hurts another group) and just merge the thing. Fortunately just that (more or less) happened on the original PR, though of course there was a discussion afterwards 😝

[parir]

s/Friedrich Nietzche/Simone/

[lambda]

cc @sferik

[oli-obk]

extra space after the .

[sferik]

s/Whean/When/

[killercup]

This is great! I understand why it seemed nice at first to quote a popular paper, but I believe this change is far better.

I've been meaning to make a PR to add some subheadings to divide this chapter into smaller sections. I'll hold that off until this was merged. (Unless you want to do that @lambda, then, by all means, go ahead :))

[sferik]

I’ll echo the comments by @mbrubeck in the internals thread on diversity on the governance teams.

For anyone who is specifically troubled by the choice of female pronouns for characters in this example, this is a good chance to empathize with people who feel the same about the constant use of male pronouns in similar situations. Consider whether you spend as much effort noticing and correcting "default male" language you encounter. Or, if you could live with the original pronouns in this text, consider whether you can also make peace with the new ones.

I also object to the omnibus fashion in which this change is being proposed. In my view, the decision about on whether to use chopsticks instead of forks, which I may agree is an improvement, is completely orthogonal to the decision to use neutral pronouns. If this was code instead of documentation, I suspect the maintainers would ask that each change you’re proposing be split into a separate pull request so that they could be considered independently.

[Manishearth]

I also object to the omnibus fashion in which this change is being proposed.

FWIW on Rust we're generally okay with omnibus PRs and only split things up if they're really large or there's a lot of discussion going on, especially for doc changes (seen a lot of "fixes" doc PRs that do a bunch of things). Some examples that come to mind.

[sferik]

FWIW on Rust we're generally okay with omnibus PRs and only split things up if they're really large or there's a lot of discussion going on, especially for doc changes (seen a lot of "fixes" doc PRs that do a bunch of things).

Thanks for explaining that. I’m quite new to the Rust community and am happy to adapt to the local standards and customs. I withdraw my objection that this should be split into multiple pull requests. Still, I hope that each individual aspect of this pull request will be considered independently.

Here is how I would vote on each aspect of this pull request, as I see it:

👍 s/fork/chopstick/g
👍 s/spaghetti/rice/g
👍 Replacing exclusively Western European male philosophers with a more diverse group.
👎 Replacing feminine pronouns with neutral pronouns.
👍 Removing trailing whitespace.

[Manishearth]

Personally ambivalent on the fork/spaghetti stuff. +1 for the philosopher replacement, -1 for the pronouns, +1 for the ws fixes.

Would prefer a solution for the "misquote" issue with the original change but am okay with the status quo on that.

[huonw]

@sferik, the author has stated (#25640 (comment)) they're happy to switch and is generally being clearly proactive in improving this text, so I implore you to make it clear you're working with them rather than setting up an adversarial atmosphere as #25640 (comment) does did. (Unless of course you truly think this PR was written in poor/misinformed faith, but even so working together is better.)

In any case, I'm very much in favour of landing this PR, with feminine pronouns and philosophers.

[sferik]

@huonw I have edited my comment, removing the paragraph that made it seem adversarial.

To be clear, I believe this pull request was opened in good faith and with good intentions.

If it is updated to use feminine pronouns, I have no remaining objections.

[pnkfelix]

@sferik are you saying you want the PR to be updated to say e.g. "on his or her left" instead of "on their left", in the cases where it is referring to philosopher's via numeric indices ("Philosopher 1", "Philosopher 2", etc)?

Or are you saying that you want the PR to be updated to say "on her left" in all cases? (Keeping in mind that the table at this point consists of, I think, two men and three women...)

[lambda]

@sferik Most of the reason I chose singular "they" was for consistency, since that's what the rest of the article uses. Changing to using feminine pronouns for the philosophers requires a bit more work to go through the whole thing, and make sure I'm only replacing uses of singular "they" as opposed to uses in which it's actually used in aggregate. I think I did that appropriately in my most recent version of the branch, please give it a look.

@pnkfelix I am pretty sure that @sferik meant that "she" should be used any place in which an unspecified philosopher was being discussed. When referring to a non-specific person, every choice of pronoun will be marked on at least one feature; if you use either "he" or "she", you are not matching the neutral gender inherent to the nature of an unspecified person, but you're matching on the features of number and animacy. If you use "they", you are matching the neutral gender and animacy, but not number. If you use "it", you are matching on number and gender, but not animacy. Not matching on animacy is generally considered to be ungrammatical, but any of the three other options are grammatical, and they each are marked on one feature, so there is no particularly good reason, grammatically, to choose one over the other; it is a stylistic choice. "He or she" covers both linguistic possibilities, but it is quite clumsy to work into many sentences.

Much prescriptive style advice over the past couple of centuries has preferred "he", even though singular "they" has been widely used for a long time. To help overcome many years of presuming a particular gender, then, it can be valuable to choose "she" instead, to balance out the many years of use of "he" which may lead to cognitive biases in that direction.

For that reason, I've updated it to use "she" for the cases in which we're discussing an unspecified philosopher.

[pnkfelix]

My personal inclination would be to either (1.) use all female philosophers and then use "she" everywhere, or (2.) use "he or she".

I acknowledge that (2) is clumsy, but its use is limited here, and for a reader like me, if you use "she" alone in a context where both men and women are present in the group, then I am going to infer based on the use of gender that somehow we had chosen a particular subject for the context, and then waste time as a reader attempting to figure out which subject I am suppose to have inferred.

"He or she" makes it abundantly clear that there is no such inference to be made.

[sferik]

@mbrubeck Thank you for moderating. As far as I can tell, the @marxk account was just created for the explicit purpose of trolling this thread.

[nox]

What about quoting the original philosophers dinner's problem verbatim with its warts and all, and commenting on how Hoare intended it as a satire of philosophers, which presumably was a domain dominated by white males in the West in the past?

Note that the alphabets of the philosophers are mutually disjoint. There is no event in which they participate jointly, so there is no way whatsoever in which they can interact or communicate with each other—a realistic reflection of the behaviour of philosophers in those days.

[nagisa]

The version of this problem I remember concerned one spoon and one fork, which actually makes some sense if you are eating spaghetti, so there’s that.

[nox]

Actually, my previous quote plus the fact that they are named PHIL0, PHIL1, PHIL2, PHIL3 and PHIL4 adds to the satire that Hoare intended them as ludicrous males that can't communicate among themselves.

We could go further and reinterpret the fact that they don't want to make any of them special (taking the fork whatever the state of the other philosophers) as a critic of the distaste of positive discrimination to fix problems among the intellectual scene.

A good writer could probably make a very fun commentary out of it, given enough time.

[steveklabnik]

@lambda , this pull request introduces three regressions that I'm concerned about:

1. Part of my goals to make Rust more inclusive is to allude to and include resources for those who are not traditionally systems programmers and those without a college CS education. I included the reference to Hoare because I originally planned on implementing this example with channels, but due to the 1.0 rush, ended up switching to Mutexes at the last minute. But including this seminal paper, and eventually adding a channels implementation, remains very important to me.
2. Switching the philosophers to ancient Greeks has its own set of problems. I chose the philosophers I did as a subtle nudge towards a body of work that more programmers should be aware of generally.
3. Using he/she still perpetuates a gender binary, which is exclusionary to trans folk to some degree. This was brought up in another ticket, and frankly, is part of why I suggested to use they in the original PR, but with all the heartbreak over this last week, I just didn't have it in me to argue. Generally speaking, we use 'they' in documentation when referring to pronouns, if not, that's a bug, as I said in the original ticket.

So, what I suggest is this:

1. Leave Hoare in.
2. Change all pronouns to they, and use [they] in the quotation.
3. Swap Spinoza for Judith Butler.

[sferik]

Part of my goals to make Rust more inclusive is to allude to and include resources for those who are not traditionally systems programmers. I included the reference to Hoare because I originally planned on implementing this example with channels, but due to the 1.0 rush, ended up switching to Mutexes at the last minute. But including this seminal paper, and eventually adding a channels implementation, remains very important to me.

It should be possible to implement the example with channels and reference the paper by Hoare without quoting from it directly. These references could be added back without changing the wording of the example.

Switching the philosophers to ancient Greeks has its own set of problems. I chose the philosophers I did as a subtle nudge towards a body of work that more programmers should be aware of generally.

Assuming Lao refers to Laozi, he was not Greek.

Assuming Simone refers to Simone de Beauvoir (or possibly Simone Weil), she was neither Greek nor ancient.

And assuming Audre refers to Audre Lorde, again, neither Green nor ancient.

If the goal is to expose programmers to new ideas, I’d speculate that the work for Audre Lorde is much less widely known than the philosophers in your original example.

Using he/she still perpetuates a gender binary, which is exclusionary to trans folk to some degree. This was brought up in another ticket, and frankly, is part of why I suggested to use they in the original PR, but with all the heartbreak over this last week, I just didn't have it in me to argue. Generally speaking, we use 'they' in documentation when referring to pronouns, if not, that's a bug, as I said in the original ticket.

I don’t believe the use of “she” or “her” is broadly exclusionary of transgender people, many of who self-identify as women and prefer being referred to with these pronouns. Of course, this is not true of all transgender people, and more broadly, genderqueer people. My intent was not to be exclusive of any group.

I have some points to make in response to #25650 but I have refrained from commenting there, since @pnkfelix requested that the conversation be moved to the thread on internals. I look forward to discussing it further on that thread.

[lambda]

1. Part of my goals to make Rust more inclusive is to allude to and include resources for those who are not traditionally systems programmers and those without a college CS education.

I am 100% on board with this goal.

I included the reference to Hoare because I originally planned on implementing this example with channels, but due to the 1.0 rush, ended up switching to Mutexes at the last minute. But including this seminal paper, and eventually adding a channels implementation, remains very important to me.

That is something that I didn't know when submitting this PR. As it is, it seemed like an odd choice for inclusion given the use of Mutexes rather than channels to solve the problem.

1. Switching the philosophers to ancient Greeks has its own set of problems. I chose the philosophers I did as a subtle nudge towards a body of work that more programmers should be aware of generally.

Only two of my set of five philosophers were Greek, Plato and Hypatia. Beyond that there was Lao (Tzu; perhaps I should have used the Laozi form instead to make it more clear) of Tao Te Ching fame, Simone (de Beauvoir), and Audre (Lorde). I chose to use a single name for each of them for symmetry with the Greek names and to make it a little bit more casual, but they were also chosen as subtle nudges towards things that programmers might need to know a little more about. I appreciate the reasons for your choice of philosophers, but it is a bit homogenous.

1. Using he/she still perpetuates a gender binary, which is exclusionary to trans folk to some degree. This was brought up in another ticket, and frankly, is part of why I suggested to use they in the original PR, but with all the heartbreak over this last week, I just didn't have it in me to argue. Generally speaking, we use 'they' in documentation when referring to pronouns, if not, that's a bug, as I said in the original ticket.

Using "they" is generally my default mode of writing as well, and I think it's a good standard to have; in fact, that's what I used in my original pull request (perhaps I should push all of the older versions of the branch as separate branches so you can see them, I've updated it several times in response to comments).

However, I am quite sensitive to the issues brought up by @sferik and @skade; that there's an overwhelming gender imbalance in this project as a whole, and in a lot of language surrounding existing discussion of this topic. Correctly or incorrectly, people have seen Rust as "that kind" of open source project, based on looking at the team roster, and it doesn't get much better if you look at the contributor list as a whole. I do think that that can warrant at least a little bit of pressure to try and tip the scale in the other direction. It's a small thing, and only one of many, but I do think that specifically including at least a few explicitly feminine pronouns in the documentation can help make it feel a little bit more inclusive.

Many trans people participate in the gender binary, preferring masculine or feminine pronouns. While there are some non-binary, neutrois, gender neutral, and agender folks out there, they are not, to my knowledge, under-represented in the way that 50% of the population, perhaps 20% of software developers, is represented as only single digit percentage of contributors, at least based on my reading of the contributors page and recent contributors lists for releases (and basing my judgement on gendered names in handles, gendered real names, and gender presentation in icons). And I think that we have plenty of documentation using the neutral "they"; this ticket is only touching a single example of five particular fictionalized philosophers.

So, what I suggest is this:

1. Leave Hoare in.
2. Change all pronouns to they, and use [they] in the quotation.
3. Swap Spinoza for Judith Butler.

That would be easing up a little bit on the all-male language in the quote, and all-male example. I don't think that it would be applying any pressure against the weight of the current demographics and historical use of language in this area, which was part of the point of my pull request (and @sferik's).

I don't believe that the language in documentation is the most important way of putting pressure on the other end of the scale; it is but one tool of many. It can help in its own way, but it's probably not worth spending too much more time on.

I'll push one last update to change Lao to Laozi to make it a little more clear; if you would prefer to take the steps you listed, we should probably just close this pull request and you can apply that change, it's far enough from the spirit of this PR that it would be better done as a new one.

[withoutboats]

I'm having difficulty following this conversation in its particulars, but it seems to me that in some comments a distinction has not been made between two distinct concepts: a) a pronoun referring to a person of unknown gender, b) a pronoun referring to any person who is a member of a set of people for whom at least some genders are known. It seems appropriate to me to use they or she to designate a) for the reasons that have been stated above, and I think reasonable people can differ about which is preferable. However, it would be inappropriate to use she in a case of b) in which the set includes people with known non-she pronouns. Put in terms of types, she is an appropriate designation for the pronoun type _, but not for the pronoun type T | U.

I would also strongly suggest singular they over he/she or he or she or (s)he or any similar conjoining of he and she, even in referring to a set whose members' pronouns are all known to be he or she, because of the history of erasure of nonbinary gender identities.

While there are some non-binary, neutrois, gender neutral, and agender folks out there, they are not, to my knowledge, under-represented in the way that 50% of the population, perhaps 20% of software developers, is represented as only single digit percentage of contributors

Maybe do not try to determine a hierarchy of oppressions. Many of the factors which cause the underrepresentation you describe effect many people with nonbinary gender identities as well as women.

[bstrie]

Considering that these threads have been linked from 4chan, I'm restricting comments here for the time being. Please take subsequent discussion to IRC or the discourse forums, where granular moderation is actually possible.
https://internals.rust-lang.org/t/pronoun-policy/2111

[Edit: Added link. ~Manish]

[steveklabnik]

@lambda, first of all, thank you for this PR, and for that last comment.

While I can see the value of positive discrimination, and am generally in favor, in this case, there are a few different things at play. While it's true that there are many strategies for dealing with pronouns, singular 'they' has a number of advantages. It's maximally inclusive in terms of groups, it's the most accurate in that it leaves the gender undefined, and it's simple to enforce as a rule. Choosing 'she' as a means of positive discrimination has some merit, for sure, but I think that it introduces procedural drawbacks, as well as focuses on one aspect of diversity over all others, which isn't inherently bad, but everything comes at a cost, you know?

Given all of these different factors, sticking with they as the proper pronoun to use in these situations feels like the right call to me.

if you would prefer to take the steps you listed, we should probably just close this pull request and you can apply that change, it's far enough from the spirit of this PR that it would be better done as a new one.

If that's what you'd like, sure. I plan on doing the following in the next few minutes:

1. Double-checking that all of our documentation uses 'they'
2. moving this quote to use [they].
3. changing up the gender balance of said philosophers
4. Later, adding a channels-based implementation to strengthen the connection between the paper and the text itself.

This discussion has been hard, but thank you all so much for caring about diversity in Rust. It's really important to me that everyone that wants to use Rust feels able to do so.