It's REALLY weird to me when people in software mine research papers for their content and say "researchers" instead of naming the scientists who actually did the work they're using. We're human beings and our work is our livelihood (at a fraction of yours I might add). Name us.
Blessed for the community around me that has this value, side eye at the content engine that doesn't.
Oh and we change institutions!! Our work is very very much a human-led affair. Funny industry quirk I've noticed is people naming the university brand like "MIT" doesn't produce a study, a lab and a scientist does. Name us!
@grimalkina that’s good to know. a lot of times industry works the opposite way, the company allows you to do project X, and a lot of times the individuals aren’t even named
@kellogh happy to share about it!! Better to think of scientists like "authors," like book writers! Even industry scientists, if they publish, still lead and author work and should be given that credit 🤗. For many of us it is the core of our reputation and when we cite each other by name we not only give credit, we also boost our own work because we show the communities of science and innovation that we're in together and that our work is stronger for that!
Yes, lots of people are used to the idea that science comes from institutions. It's even taught as part of basic information literacy: does this come from a respectable large organization, or from one person (probably a crackpot)?
Even a classmate in 1st-year library science grad school was surprised to learn that a study by SoandSo, Ph.D, could be just as reliable as a study from Institute X.
She'd been in 'industry' before, so maybe that was her previous training.
@Kathmandu@kellogh I mean it's not like affiliation and institutions signal nothing, of course they signal that you meet the criteria of the institution which frankly is often important. But you know it just doesn't accurately reflect the work & credit involved.
@grimalkina@Kathmandu@kellogh I like to know if something is from Stanford because I have almost zero trust in that institution. Harvard is also low.
But you are absolutely right.
@grimalkina I might need to audit my own writing to make sure I have not done this! I wonder if this maps on to the cultural norm that sw devs just expect to grab a bunch of dependencies and never much bother with even reviewing the licenses, let alone crediting the authors by name. We are worse to each other than to researchers!
@glyph I totally can see that! worth thinking about for sure, I've heard so many developers express a lot of sadness at the feeling that their intellectual work simply vanishes away from them and I feel like ownership and authorship is something all folks need in some way!
@grimalkina this has caused me to rack my brain for any concept of citation or credit in open source culture and like… wow we just don't really have it anywhere, do we. A SBOM manifest is nothing at all like an acknowledgements section or a bibliography. Quite a blind spot!
@glyph@grimalkina We used to spend a lot of love and care on the GNOME about box; Mozilla built a literal monument out of theirs.
Wikipedia once also had a team experimenting on how to joyfully and engagingly surface "this page made by humans like you" but it sadly never got anywhere—which I agree is more typical than the occasional human-facing about box.
@luis_in_brief@grimalkina even about boxes in themselves are a bit of an out-of-the-way nook to stash credit for things. People don't aspire to roll credits on an app the same way they do in a video game. But I'm thinking of just within the artifacts themselves, libraries which depend on other libraries, don't give credit in this way. And did those about boxes include credits for library authors unaffiliated with gnome/mozilla, or was it just for folks who worked on the apps themselves?
@glyph@grimalkina for GNOME and Mozilla it was a fairly broad trawl (including docs and translators, and for GNOME I'm pretty sure it included gtk), though obviously not all the way down the stack.
Libraries are a real UI challenge. More broadly, our systems are simply so huge that surfacing "credit" in the academic sense requires a great deal of creativity, to split the difference between comprehensiveness (which inevitably is huge) and selection (which inevitably has... problems).
@glyph@grimalkina I can't find it now, but for a keynote (and at least for a while, in the software?) GitHub had some nifty visualizations of "this is who built it" but it was much more about "giving you a sense that this created built on the shoulders of a huge number of real people" and less "giving you a deep sense of the identity of each specific person".
@luis_in_brief@glyph@grimalkina yeah it is really hard to communicate the sheer size of the dependency tree and authorship for software, so we simply stopped doing it.
It is also partially how we got into the current "supply chain" discourse. A lot of the "thought leadership" on it has no realization of the sheer size of it, they only see what skimmed up. So we get discourse highly separated from the reality far too regularly.
@Di4na@glyph@grimalkina yeah, that’s definitely true. It’d be a better world if every CTO+CISO knew basic facts about their dependency tree: how many packages, median # of maintainers, MTBF.
(Now I’m wondering what that diagnostic would look like. Very different from an SBOM or license scan report.)
@Di4na@luis_in_brief@grimalkina this is exactly not the problem I am talking about here :-). Like that’s a serious problem as well, but what I am talking about is a cultural and social norm of giving credit to people who do important work, in a way analogous to film credits or academic citations, which typically include people’s names and not just product/project/brand identifiers
@glyph@Di4na@luis_in_brief with appreciation for the many implications toward eg mapping technology brittleness and oversight, I was having the same thought that I was thinking about developers themselves having a lifetime view of their own work and sense of legacy and their own personal records of effort but these convos always turn into being about mapping dependencies and the tech 😊
@grimalkina@Di4na@luis_in_brief I can't really blame anybody for veering in that direction, it is the looming colossus on the horizon of any discussion of public software infrastructure, visible from every vantage point
@glyph@Di4na@luis_in_brief intersecting and overlapping things for sure and joint solutions perhaps too! No blame for the interesting/important points here I just so often struggle to get space for the other conversation! (Not with present company but in the industry)
@grimalkina@glyph also, I came into the discussion sideways as one does on social after one has been out 🍹ing. Sorry!
You might look at CZI’s work—they support some work on a citation-centric approach, driven by the intuition that one of the reasons scientific FOSS is under-supported is that academic-track devs can’t get “credit” in the currency of the realm (citations). Related, same for data: https://makedatacount.org/data-citation/
@grimalkina@glyph@dartar (that’s still a pretty sterile view, and suffers from all the problems of the citation-industrial-complex, but it does at least center the challenge on creators and their own need for understanding, which I think is what you’re also trying to get at?)
@luis_in_brief@glyph@dartar thank you and agreed! I know some folks who went over to CZI 🤗 my wife actually does neural data science work/has much contact with large open datasets in neuro/ and is more in this world than I am but have seen a lot because of that! I think some of these scientific+software approaches are fascinating/very heartfelt/a place we can all learn from each other!
@grimalkina@glyph@dartar ironically, I come at it in large part from a completely failed attempt: licensing. Essentially every open license has attribution-like requirements, and yet even my meager examples are not driven by licensing.
The conflict drives me bonkers: is attribution a strongly held-preference of developers (as you’d guess if all you had was the licenses) or something no one cares about (as you’d guess from the outputs)?
@luis_in_brief@glyph@dartar my stance as a psychologist for developers is that developers' preferences are highly divided on a lot of things!!! A lot of conflicting norms in this field (e.g., "everyone can learn to be a developer" but also "only some of us are genuises"). I also wonder just about awareness at all of the concerns here
Preferences can be highly malleable, perhaps a useful framing of a question is what causes this preference for attribution to emerge strongly for developers?
@grimalkina@luis_in_brief@dartar so I am not sure if @r0ml has posted about this online, but one thing he has said about licensing in his talks is that licensing and litigation are not very effective or appropriate mechanisms to deal with enforcement of plagiarism & attribution. Academia doesn’t do it that way, film doesn’t do it that way. And thanks to Hbomberguy’s recent research in the lead-in to his plagiarism video, we know that it is phenomenally hard to use those tools in that way
@grimalkina@luis_in_brief@dartar Harlan Ellison’s exemplar Brillo case predates the existence of free software licensing, and it was way more cut and dried than a nuance of license enforcement, it was full-on criminal infringement and yet nobody else seems to have pulled off a similarly prominent case since.
@glyph@grimalkina would love to read more about that if you’ve written it up, @r0ml. At least at the highest level, I strongly agree, which is why I’ve advocated taking it out of the licenses and doing it differently.
(But to your “it’s the colossus” point, FOSS attribution is a “quantity has become a quality” situation—our norms are from the early days of film with a dozen names in the credits, but a realistic practice these days would require minutes of very, very tiny scrolling font.)
@grimalkina@glyph great question. I wonder if CC has any old writing on this - they (we!?) had a no-attribution option in the early years, but something like 98% of users chose an attribution-based variant, so it was removed. (It was then resurrected, in different form, as CC0.)
I have no idea what, if any, research CC did into that at the time - maybe @mlinksva knows?
I only vaugely recall, but I think a big part of the rationale was simplification, including for CC -- at the time it was just starting to "port" licenses to as many jurisdictions as possible, and 5 fewer was good. Of course the better simplification later was to stop porting. I was no doubt enthusiastic about getting rid of plain SA for being incompatible with BY-SA but I doubt anyone else cared.
@mlinksva@luis_in_brief@grimalkina@glyph something to keep in mind is that there is probably a natural attraction, if presented with the choice, to say "oh attribution would be nice".
But that does not mean there is a loss feeling if the attribution does not happen.
I know for my own work, attribution would maybe make me happy a bit but in practice... Meh? Who cares?
@luis_in_brief@grimalkina it's a significant challenge to be sure, but I'm… lightly embarrassed that I haven't thought about this in this way before, even having spent a truly disproportionate amount of my life thinking about the motivations of and benefits to open source developers. Citations at scale in the academic world are also a complex and comprehensive problem but there is a lot of energy spent maintaining the systems which support them
Add comment