Item — HN Reader

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.