It’s time (past time, really) for a pledge drive! If you appreciate this blog and its unique perspective, please consider donating by clicking the button below.
You can also donate by Venmo at David-Futrelle-1.
There’s no other blog quite like this one and your donations are what keep it alive. WHTM is ad-free, and I don’t pester you with popups asking for donations every time you visit. (I try to limit my pestering to these pledge drives.) There’s no mysterious billionaire keeping me afloat. This is a one dude, two cat operation.
I know this is a difficult time to ask for donations; many of you I am sure are as strapped as I am at the moment. But if you can afford to donate, please do. And if you can donate a little extra to make up for those who can’t donate now, that would be even better. You don’t know how much I appreciate it.
And if you’re one of those who donates monthly or has donated a lump sum since the last pledge drive, thank you so much!
My financial situation is, I’m sad to report, pretty precarious. I don’t have the money to cover either my rent or my insurance payment at the end of the month. And I’m still struggling with medical bills. I had to cover several thousand dollars worth of emergency dental work this spring and at the moment I don’t have the money to actually get the job finished. So every little bit you can spare would help a great deal.
Thanks in advance for your support!r
–David Futrelle
Cat pics?
I’m very sorry to hear that. I really want to see cat pics but I won’t hold my donation hostage for them. Cat pics, tho!
Need to hit the bank tomorrow anyways, so I’ll remember to shift a bit over to use.
You’ve given me hours of reading – and a fair amount of learning. Seems fair enough to chip in.
Mind … cat pics are always welcome. 😛
What? You need to talk to George and get that check.
I’m making a donation after I post this comment.
I’m again going to recommend David to open up a Patreon to have a more of a steady income.
Donated, also seconding Catalpa’s Patreon idea.
Hey everyone can I have your attention? If you have any extra to spare could you help our good friend Paradoxy? They’re in need of some cash for rent.
https://twitter.com/ParadoxyIntent/status/1288540268426416128
On an unrelated note, anyone heard of “genderify”? Awesome premise… stick a name or email address into their magical web API, and it’ll tell you whether the person it belongs to is male or female! Using the power of AI! It worked about as well as you might expect. It seems to have gone from launch to overwhelming mockery to shutting up shop in about 48 hours.
https://www.theverge.com/2020/7/29/21346310/ai-service-gender-verification-identification-genderify
I’ve seen programs that will predict someone’s gender based on a writing sample, with more words giving you a better prediction, but never down to just a name or email address. That’s pretty arrogant! LOL
Donated, and might give a bit more but need to think about it. I hope you receive all you need, David.
@Pie
I have to admit; I did a double take when I read that. I’d somewhat understand an AI that tries to gender photos, even though that is still also problematic it at least makes a modicum of sense as an idea. But email addresses and usernames? Most people I know, myself included, have emails that don’t really look like names and don’t give any indications of gender. Same with Twitter handles or other usernames.
Reminds me a bit of how some TERF created a social media platform that requires users to upload photos which were then checked by AI to see if they were cis female. In addition to being illegal under UK law (where it was based), the app appears to mostly check if someone is white and feminine.
And the genderify disaster also remind that nobody ever know what a neural network actually learn. One of the strong example was how putting “Dr.” in front of a name made the AI see it as male. That’s a good example of how neural networks can appear to work and be based on obviously bogus rules.
Similarly, somebody tried to use neural network to detect criminal. It “worked” in practice by seeking if the suspect was non white and if he was poor. It’s not just that it reflect the bias of its trainer, it’s also that nobody can double check if the method used make sense.
“pledge”. always makes me think of ma dad saying “son, you don’t OWE allegiance to a flag”
@PoM
Out of curiosity, I just checked out one of the programs that predicts gender based on writing samples and pasted in some things I’d written. It gendered me correctly as female.
That being said, I still find the premise sketchy and I don’t like that it relies on binary ideas of gender or the idea that men or women should write or sound a certain way.
@Naglfar
Back when I was on Livejournal writing fanfiction, my friends and I all put our writing through one of the bots. It was correct about 80% of the time, not counting myself where any gendering is both correct and incorrect.
The word “should” implies that the bot is prescriptive, but it isn’t. It’s just observing that men and women do, factually, have different-sounding written voices and making predictions based on those factual patterns. There’s nothing in the bots that says you SHOULD write in a particular way, but that you just do.
I’ve done individual donations before, but I’ve upped to donate monthly. I hope it helps.
While I have the resources I’m trying to help others.
Your work has been very valuable to me and I read it everyday.
Those “let the machine determine if this person is actually [x]” are stupid at best, and biased and harmful at worst.
That said, it would be kind of cool to have an extension or something that would scan for preferred pronouns in a person’s bio and plug in the correct ones if you mentioned the person. I’ve accidentally misgendered someone once or twice because I wasn’t paying attention to their bio and that’s always embarrassing.
Out of curiosity, I ran a number of TERF blog posts by women through the bot and it has declared them all to be male. OTOH, it consistently genders me as a woman. I can just imagine the rage they’d have if they realized this.
To be clear, I do not like the bot and do not endorse it. I just find it a bit interesting how a group of cis women obsessed with always being able to tell trans* women apart gets misgendered by the bot.
I’m using this bot BTW.
Got curious enough to plug the first chapters of two novels into that bot.
The one with the male MC registers as weak male, with slightly over 52% “male gendered” words.
The one with the female MC registers as weak female, with slightly under 54% “female gendered” words.
(side note, I find the note that weak male may indicate European to be hilarious)
But we’re talking almost 50/50 splits of gendered words in each case. Which is good. I like balance. 😛
@Naglfar:
Due to sex discrimination laws? Do those even apply to noncommercial private spaces?
Deliberately excluding non-whites, or an accident because they only trained it on white faces or something?
Of course, this would be ridiculously easy to defeat even if it worked as advertised. Just upload a photo of, say, Daniela Pestova. I don’t think they get much more feminine or white than her. Bam, infiltration achieved, now we can spy on the enemy camp. 🙂
@Ohlmann:
https://www.goodreads.com/book/show/66479.Starfish
@Surplus
In the UK, these things are governed by the Equalities Act, which I am told (but could be wrong, please correct me) says that single-sex spaces can only legally exclude trans* people (which this app claims to do) if there is a justification that stands in a court of law. There is no proportionate means justification that would support this app.
I don’t know if it’s deliberate (but might be), I just know that the vast majority of facial recognition AIs are mainly trained on white faces and most of the cultural signifiers of femininity in European culture have been designed around whiteness to exclude non-white women. Of course, this also excludes GNC women of all races, which might be a feature and not a bug for them.
I actually don’t know how they prevent this. And since it appears a lot of TERFs online have/are sockpuppet accounts, this probably has already happened.
Honestly, I don’t think there is much of a point to infiltrating it. Infiltrating other TERF social media sites like Spinster or Glindr is easy but yields virtually no useful information, it’s just a lot of people obsessing about trans* people.
I did the gender bot @Naglfar linked to. I’m male, apparently. And they’re really sure, too, like 75% dude.
(I’m not male.)
Also, it said the genre of the sample is “informal” so I can’t take this seriously at all. (I put a couple of pages of the very formal academic book I am writing. It has footnotes, for heaven’s sake.)
@Susan
The two boxes outlined in red under the long horizontal total words box outlined in red appear to give two different analyses, one assuming formal writing and the other assuming informal writing. That’s what it’s shown for all the samples I’ve put through it, at any rate.
@Kinda, Aha! I see that now. Thanks. It’s still telling me I am male. By a lot.
I just ran a bunch of the Warren Farrell quotes from an earlier post here, and it is convinced he is female. It also appears to believe Valerie Solanas and Andrea Dworkin are male. In conjunction with other observation reported here, it seems this bot doesn’t work very well.
I also fed it text from another AI, and that AI consistently is gendered as male. Maybe the other AI’s training data was predominantly by men?
OT: I was locked out of this site for several hours this evening, without cause or any sort of warning. During this interval I was forbidden to even read anything here. This seems to require an explanation. Why was I subjected to this atrocious treatment? I am unaware of having done anything controversial recently to warrant any sort of a “time-out”.
I challenge the person responsible to step forward, admit responsibility, and explain their actions.