Following our Code of Conduct the project aims to be a space where people are considerate in natural language communication:
There are terms in computing that were probably considered benign when introduced but are uncomfortable to some. The project aims to de-emphasize such terms in favor of alternatives that are at least as expressive - but often manage to be even more descriptive.
A common thread in discussions was that the project merely follows some fad, or that this is a “political correctness” measure, designed to please one particular “team”. While the project doesn’t exist in a vacuum and so there are outside influences on project members, the proposal wasn’t made with the purpose of demonstrating allegiance to any given cause - except one:
There are people who feel uncomfortable with some terms being used, especially when that use takes them out of their grave context (e.g. slave when discussing slavery) and applies them to a rather benign topic (e.g. coordination of multiple technical systems), taking away the gravity of the term.
That gets especially jarring when people aren’t exposed to such terms in abstract sociological discussions but when they stand for real issues they encountered.
When having to choose between using a well-established term that affects people negatively who could otherwise contribute more happily and undisturbed or an alternative just-as-good term that doesn’t, the decision should be simple.
The other major point of contention is that such decisions are a token gesture that doesn’t change anything. It’s true: No slave is freed because coreboot rejects the use of the word.
coreboot is ambitious enough as-is, in that the project offers an alternative approach to firmware, sometimes against the vested interests (and deep pockets) of the leaders of a multi-billion dollar industry. Changing the preferred vocabulary isn’t another attempt at changing the world, it’s one thing we do to try to make coreboot (and coreboot only) a comfortable environment for everybody.
For everybody, but with a qualifier: We have certain community etiquette, and we define some behavior we don’t accept in our community, both detailed in the Code of Conduct.
Other than that, we’re trying to accommodate people: The CoC lays out that language should be interpreted as friendly by default, and to be graceful in light of accidents. This also applies to the use of terms that the project tries to avoid: The consequence of the use of such terms (unless obviously employed to provoke a reaction - in that case, please contact the arbitration team as outlined in the Code of Conduct) should be a friendly reminder. The project is slow to sanction and that won’t change just because the wrong kind of words is used.
Interfacing with the world¶
The project doesn’t exist in a vacuum, and that also applies to the choice of words made by other initiatives in low-level technology. When JEDEC calls the participants of a SPI transaction “master” and “slave”, there’s little we can do about that. We could decide to use different terms, but that wouldn’t make things easier but harder, because such a deliberate departure means that the original terms (and their original use) gain lots of visibility every time (so there’s no practical advantage) while adding confusion, and therefore even more attention, to that situation.
Sometimes there are abbreviations that can be used as substitutes, and in that case the recommendation is to do that.
As terms that we found to be best avoided are replaced in such initiatives, we can follow up. Members of the community with leverage in such organizations are encouraged to raise the concern there.
Dealing with uses¶
There are existing uses in our documentation and code. When we decide to retire a term that doesn’t mean that everybody is supposed to stop doing whatever they’re doing and spend their time on purging terms. Instead, ongoing development should look for alternatives (and so this could come up in review).
People can go through existing code and docs and sort out older instances, and while that’s encouraged it’s no “stop the world” event. Changes in flight in review may still be merged with such terms intact, but if there’s more work required for other reasons, we’d encourage moving away from such terms.
This document has a section on retired terms, presenting the rationale as well as alternative terms that could be used instead. The main goal is to be expressive: There’s no point in just picking any alternative term, choose something that explains the purpose well.
As mentioned, missteps will happen. Point them out, but assume no ill intent for as long as you can manage.
Discussing words to remove from active use¶
There ought to be some process when terminology is brought up as a negative to avoid. Do not to tell people that “they’re feeling wrong” when they have a negative reaction to certain terms, but also try to avoid being offended for the sake of others.
When bringing up a term, on the project’s mailing list or, if you don’t feel safe doing that, by contacting the arbitration team, explain what’s wrong with the term and offer alternatives for uses within coreboot.
With a term under discussion, see if there’s particular value for us to continue using the term (maybe in limited situations, like continuing to use “slave” in SPI related code).
Once the arbitration team considers the topic discussed completely and found a consensus, it will present a decision in a leadership meeting. It should explain why a term should or should not be used and in the latter case offer alternatives. These decisions shall then be added to this document.
Replacing this term for something else had the highest approval rating in early discussions, so it seems pretty universally considered a bad choice and therefore should be avoided where possible.
An exception is made where it’s a term used in current standards and data sheets: Trying to “hide” the term in such cases only puts a spotlight on it every time code and data sheet are compared.
Alternatives: subordinate, secondary, follower