back
▲ 5 points

Show HN: A tool to find books outside gender echo chambers

abooklikefoo.com
by padolsey·3y ago·4 comments·view on hn ↗
Hi HN. I made this mostly for my own curiosity. Many of us end up reading books by or about people of specific genders. So it's interesting to find books that are outside that echo chamber but still mostly _inside_ the topical or affinity area we are most engaged in. Doing this, it helps us to each expand our reading while elevating underrepresented authors, topics and character-types in the books we read.

For myself, I found it most interesting to find more female writers in the topic area of economics and technology, where I've experienced a high concentration of male writers.

From the explainer: . Imagine sitting in a library of limited size, tailored perfectly to your tastes. If we look at all the books in this library, the genders of their authors, and the genders of characters within the books, then we can find your "literary gender bias". That's what this tool attempts to do. Giving the tool 3-5 books can serve as the seeds of this hypothetical library. From these, we can then extrapolate, calculate and show you your gender bias, and a set of recommendations that seek to invert that bias.

4 comments
What a difficult project to do. You have to go confirm each possible author's gender before you could list them. Some percentage I'm sure would be easy to find the answer online but you wouldn't want to misgender authors and either wrongfully include or exclude.

As such I tested it out. I search Lying by Sam Harris. Which it says is 96% male genre, fair enough it's quite the sausagefest in Sam Harris circles. It "debiased" recommendations by biasing 96% toward women. So not fixing bias but just creating more bias.

The female recommendations: Will Schwalbe, Sam Harris, Christopher Hitchens, Oliver Sacks, Richard Dawkins.

Hmm, the recommendations do seem good though. I didn't know Hitchens identified as female.

Oh gosh, you have my respect. You've gone right in with the apex challenge there; to de-bias the sam harris readership!

It's important to note that the recommendations are an *attempt* at a bias inversion while maintaining affinity to the searched title. And gender in the literary space, is, curiously, along a spectrum. So indeed it might be the case that Oliver Sacks is less "male"-biased (as an author) than the likes of Sam Harris or that Jordan P fella (ugh). I see in the recommendation set, tho, that there are some notable female authors, so I shan't consider it an utter failure :D

Worth noting as well: the authors' genders are not noted manually. From the implementation notes:

> Over 150,000 books have been given a very approximate score based on the names of the authors (compared against 80k locale-specific names) and the frequency and density of pronouns used in descriptions and in many millions of reviews of the books. This gives a very broad picture of the gender-bias space, but obviously lacks accuracy for any specific title. Nonetheless it enables us to build a general picture because books tend to be in "affinity spaces" with similar books. You can read more about how recommendations are made in a general sense on abooklikefoo.com as this uses the same backing data and hueristics.

EDIT:

> So not fixing bias but just creating more bias

I have .. opinions on this (Miles-long screed in which I navigate the topic of how best we go about creating more equitable systems where previously systemic biases have held sway for so long. /... years later after we gather to read libraries of reflections on affirmative action and positive discriminations.../ we border on agreement that perhaps it's ok to counter a 95:5 bias with 5:95 and not be too perturbed :p)

>Oh gosh, you have my respect. You've gone right in with the apex challenge there; to de-bias the sam harris readership!

To be clear we're not talking about prejudice but rather just bias. Which is certainly interesting to why Harris biases toward men, but not really seeing the problem. There is absolutely no problem in having a bias. Everyone and everything has to figure out their target audience and more often than not you will pick your own identity group.

>Worth noting as well: the authors' genders are not noted manually. From the implementation notes:

So you are actively misgendering and discriminating against trans people amongst virtually everyone else. .

>I have .. opinions on this (Miles-long screed in which I navigate the topic of how best we go about creating more equitable systems where previously systemic biases have held sway for so long. /... years later after we gather to read libraries of reflections on affirmative action and positive discriminations.../ we border on agreement that perhaps it's ok to counter a 95:5 bias with 5:95 and not be too perturbed :p)

I guess we disagree here. You flagrantly discriminate against everyone and then introduce a completely broken biased results. The way I see it you are likely to be a bad actor here.

Thanks for your input.