Timeline for Stack Overflow Survey: You're still spending too much time focusing on demographic groups, and not enough time improving site mechanics
Current License: CC BY-SA 4.0
32 events
| when toggle format | what | by | license | comment | |
|---|---|---|---|---|---|
| Aug 13, 2023 at 4:56 | comment | added | Karl Knechtel | My citation is the fact that people have skin colours, facial geometries, etc. which are strongly genetic and can be visually inspected. That a different culture (and yes, the US contains many different cultures) can have a different folk conception of race that clusters differently, doesn't make race "dubious as a concept", any more than the fact that Japanese people use the word 青 (literally blue) for many things I would consider 緑 (green) makes colour "dubious as a concept". | |
| Aug 12, 2023 at 17:38 | comment | added | Shog9 | I'm gonna say "citation needed", @Karl. In the US, where race has been codified into law in various forms for its entire history, there are loads of examples of folks "changing" their race by just... Moving to a new place and not telling anyone. That tends to be the way with societal constructs: they're only as constraining as the society that imposes them. | |
| Aug 12, 2023 at 11:06 | comment | added | Karl Knechtel | "Race is dubious as a concept" - how do you reconcile this assertion with the fact that children of two parents generally deemed to be of the same race, are overwhelmingly deemed to also be of the same race in the same societal context, even by people who don't know who the child's parents are? And that this result persists even when the parents are of different ethnicities, practice different religions etc.? | |
| Oct 9, 2020 at 16:05 | comment | added | FreeMan | @10Rep you're most welcome. There are certain phrases and somewhat awkward word orderings that usually tag someone as a non-native English speaker. Of course, they all speak/write English far, far better than I speak/write their language! A smattering of French, a bit of Hebrew, and I can count to 10 in 4 or 5 other languages, so full marks to all of them! | |
| Oct 9, 2020 at 16:01 | comment | added | 10 Rep | @FreeMan I'm flattered :). And you've actually covered my point! It's impossible to group people into ethnicity based on their level of grammar. I've seen the french stereotypical accent, but that doesn't mean I would say everyone who has different word order is not american. | |
| Oct 9, 2020 at 14:07 | comment | added | FreeMan | @10Rep - I've worked with a multitude of Indians, both in-person and offshore. Usually, I can identify an Indian background based on phraseology and word order. There is no way in the world I'd have pegged you as Indian based on that comment! Your grammar is probably better than 50%, nah... 75% of Americans! | |
| Oct 9, 2020 at 14:05 | comment | added | FreeMan | Sadly, @Shog9, my phone's keyboard will often put a space before punctuation, even though I've selected US-English. I actually care enough about how I communicate that I'll reread and proof-read my messages (except when I've made a blatant and embarrassing typo, it seems) and fix it, even if it's "just" a message to my wife who is willing to overlook my random typos, or, if she notices them, just laugh. #TIL that there is a cultural reason for this, maybe my keyboard's programmers are all French... | |
| Oct 8, 2020 at 19:10 | comment | added | Shog9 | I don't disagree at all, @10Rep. Demographics can help you understand where you might be failing a given population, but it can't as easily allow you to better serve them (a somewhat benign example that you may have seen involves browsers or websites using demographic heuristics to adjust their display or behavior - I got charged extra for a plane ticket once because the airline mistakenly thought I was in a different country...) One of many reasons why I so stridently urge care and caution with the collection and use of this data. | |
| Oct 8, 2020 at 17:23 | comment | added | 10 Rep | @Shog9 Also, I agree with most of your answer, except for the part where you say that its useful to collect ethnical data to determine their grammar. As an Indian, I have decent grammar and word usage. Certainly not the most salubrious out there, but better than some people. And if an indian user (or anyone of any other ethnicity) makes a post with poor grammar, one can't assume that it will apply to everyone of that same ethnicity. One might say it's generalizing. Not accusing you of anything, but I just wanted to put my point out there :) | |
| Oct 8, 2020 at 17:13 | comment | added | 10 Rep | @PeterMortensen I'm not going to say it's offensive, but it hasn't originated from India. And plus, how is knowing who's Indian and who's not even relevant to SO? It's kind of hurting to me, an Indian with decent grammar, for someone to say that an Indian's english skills are weird and different. I'm an Indian, so I do know this as a fact. | |
| Oct 8, 2020 at 0:45 | comment | added | Peter Mortensen | That is 99% by Indian/Pakistani writers (due to the very late adoption of later editions of the 1935 Wren & Martin textbook) | |
| Oct 6, 2020 at 17:05 | history | edited | Shog9 | CC BY-SA 4.0 |
added 532 characters in body
|
| Oct 6, 2020 at 14:51 | comment | added | Shog9 | One of my favorite examples is the space-before-punctuation thing that French writers use at the end of sentences and often carry over into English text, @Peter. A simple thing without any inherent implication apart from native style. | |
| Oct 6, 2020 at 14:47 | comment | added | Shog9 | I'm afraid this is probably true, @dalija; if we look at how the survey is structured and the few reports that were posted at the start of the year... It looks an awful lot like it's being used to calculate a loyalty score. IOW, an inscrutable metric without clear or specific meaning that can be used to justify doing - or not doing - whatever its interpreter desires. If true, that's... Not good. | |
| Oct 6, 2020 at 14:06 | comment | added | Dalija Prasnikar Mod | When it comes to company and trying to understand particular group of users and their needs focusing on broad demographic is wrong approach. It will never yield good results. If there is a particular issue that need to be solved then focus of any survey or any other analysis must be very precise and must deal with that particular problem only, leaving very little room for misinterpretation. | |
| Oct 6, 2020 at 14:01 | comment | added | Dalija Prasnikar Mod | Equality has many faces. And even when you try to nominally provide equal treatment, that does not mean that it will be fully unbiased and effects will almost certainly be different depending on the receiver. But, this is unavoidable and as long as we collectively try our best to really treat people equally and without discrimination, I think we are doing fine. | |
| Oct 6, 2020 at 9:29 | comment | added | Peter Mortensen | Can you elaborate a little bit on "default writing styles"? E.g., what would be an example of one? | |
| Oct 6, 2020 at 7:33 | comment | added | Heinzi | "...we ran into this exact problem face-first when we started rolling out International sites" - that sounds like a really interesting story. | |
| Oct 5, 2020 at 21:58 | comment | added | Shog9 | I agree with the sentiment, @Dalija - but implementing that gets into a lot of details that aren't so easy. When folks talk about "equality", they can mean a lot of different things: equal opportunity, equal treatment, equal effect... For example: if I speak English to an English-speaker, and English to a Japanese-speaker, I'm nominally providing equal treatment - but the effects aren't equal: one person hears their native tongue, another does not. That isn't necessarily unreasonable, of course - but, it does have effects, and if you're seeking to understand them then information is useful. | |
| Oct 5, 2020 at 21:34 | comment | added | Robert Harvey | Love the report card; I think the judgements in it are reasonably fair. | |
| Oct 5, 2020 at 21:26 | comment | added | Dalija Prasnikar Mod | Main problem is onboarding new users. And new users come in all shapes and colors... the only common thing they have is they are new to this site and they are not familiar with the rules. | |
| Oct 5, 2020 at 21:23 | comment | added | Dalija Prasnikar Mod | Heh... what I want to say is there is no "done well". There is not even "done better". Thinking that kind of information can help improving anything is merely an illusion. There are many things SO can do to improve the site(s). The best part of SO was always "we don't care who you are, we care about quality of your content, and last but not the least we treat everyone equally". Treating everyone equally can only be achieved if you literally treat everyone equally. The second you start treating someone differently, equality is gone. | |
| Oct 5, 2020 at 21:09 | comment | added | NoDataDumpNoContribution | I like this answer a lot because is most prominently feature that it's less the data that's a problem and more the conclusions that people draw from it. | |
| Oct 5, 2020 at 20:48 | comment | added | jpmc26 | SO no longer demonstrates care or oversight. Their record is simply appalling. Saying they need to use care and oversight is like telling a pyromaniac not to light the box of matches they're holding. SO has actively looked for justifications for advancing class grievances for well over a year now. They need to stay away from this stuff until they show they can be trusted with it. | |
| Oct 5, 2020 at 20:24 | comment | added | Shog9 | Perpetuation isn't a necessary outcome, but it is certainly a potential one, @jpmc26. Lest we forget, redlining and worse has been the predominate use of exactly this sort of demographic data in the past! That doesn't mean this dataset has to be used in such a manner, but it does underline the potential for harm and the need for care and oversight. | |
| Oct 5, 2020 at 20:23 | comment | added | jpmc26 | @Shog9 I would expect that actively grouping people based on race data can do nothing but perpetuate the historical divisions. If you focus instead on the non-racial qualities, you take an active step toward breaking down any artificial divisions. | |
| Oct 5, 2020 at 20:22 | comment | added | Shog9 | Yes, @Dalija - this is a brutally difficult undertaking. All the more reason to make sure it is being done well. | |
| Oct 5, 2020 at 20:20 | comment | added | Shog9 | Race is dubious as a concept, @jpmc26; the fact that it can correspond to such things in specific contexts comes down to how that concept has been used in the past to divide people and the social/cultural elements that have grown up as a result. IOW, it can still be useful, but you need other information (locale, age, etc) to make use of it. | |
| Oct 5, 2020 at 20:20 | comment | added | Makoto | It's not that we don't talk about the poor analysis job because it's a convenient example of poor analysis Shog; it's that we bring it up often as a reminder that poor analysis is toxic and far more detrimental to one's overarching goals than they realize. | |
| Oct 5, 2020 at 20:20 | comment | added | Dalija Prasnikar Mod | "demographic information used to identify biases in some part of the system" and here is the problem... even if you have all the data identifying the bias may be mission impossible. And SO does not even have the actual data to begin with. Only fraction of users complete surveys, you cannot even be sure they statically correctly represent all users and there is no way to validate whether demographic data they provide is true. | |
| Oct 5, 2020 at 20:18 | comment | added | jpmc26 | "...different default writing styles..." This requires assuming that race influences your writing style. If you want to ask about writing styles, develop questions that identify a person's writing style rather than assuming race has anything to do with it. | |
| Oct 5, 2020 at 20:13 | history | answered | Shog9 | CC BY-SA 4.0 |