The Unit of Caring

you gave me wings when you showed me birds

kaminiwa:

theunitofcaring:

I’ve had the chance to observe a lot of engineering hiring lately and it’s given me a lot of thoughts about diversity hiring.

Much of the screening process to hire for software jobs happens before you actually see an engineer write any code; you look at their resume and decide if they’re worth calling back, you talk to them on the phone and decide if they’re worth sending an at-home screen of some kind, you look at that and decide whether to bring them in for an interview. 

Now, obviously you can change around your standards for who you talk to and who you bring on for an on-site interview. There’s always a chance that a candidate will really impress you at the interview, even if they have a mediocre resume and aren’t impressive on a phone screen, but it’s not as likely. Maybe candidates who have pretty limited resumes and aren’t impressive on the phone have only a 5% chance of turning out, once you do an on-site, to be someone you want to hire. While candidates with a great resume who sound amazing on the phone have a 30% chance of turning out to be someone you want to hire.

Your engineering team does not want to interview twenty people to find someone hireable. They like writing software, not interviewing. If they sit through fifteen days of interviews with people who suck and aren’t hireable, they will be miserable. You will end up setting your bar higher, so that you don’t bother bringing people on for an onsite which they have only a 15% chance of passing, and you’ll inevitably pass up some great engineers you would have hired but you’ll also waste a lot less time.

So your engineering team wants more women/people of color/veterans/disabled people/etc. I feel like the narrative I’ve heard around affirmative action has been ‘either you can use it as a tiebreaker (i.e., when you have two equally qualified candidates you pick the candidate with the background you want) or you can pass up qualified candidates to pick less qualified ones or you can give up on diversity hiring altogether’.

And obviously there’s another option: you can be likelier to talk to them in the first place. You can let someone through to have an interview even if you think, based on the phone screen, there’s only a 15% chance that after the interview you’ll want to make them an offer. This does not involve hiring less qualified candidates, just talking to more people in order to miss fewer qualified ones. I think this is the most common actual diversity tradeoff involved in actual hiring, and it’s weird to me that it’s not even mentioned as a form diversity hiring can take when people are talking about the challenges of doing hiring appropriately.

(Incidentally the ‘tiebreaker’ thing is very very rarely what is going on, and pretty much never at a company like Google or Facebook. If Google interviews two great engineers they want, they’ll hire both of them, they don’t have to choose. They’re not trying to fill specific openings, they’re trying to get good engineers and place them on a team that has an opening.)

You’re modelling hiring as “a good employee has skill >= 10″, whereas most engineer hiring I know is “we want to hire the highest skill employee.”

Your system is great for identifying women who have the minimum skill necessary to work for the company. This is very different from identifying the most qualified candidate.

If Alice got a 50% on the phone screen and Bob got a 90%, then all else being equal, our available information says we should still hire Bob because he has the higher skill.

If we hire Alice instead, then we are pretty clearly focusing on diversity INSTEAD of hiring the most qualified applicant. That doesn’t mean Alice isn’t qualified, or has a skill less than 10 - it just means she’s not the highest skill in the applicant pool.

What’s more, this is going to result in a higher gender variance in work results - you’re going to get a lot of women who are right around the cutoff of 10 skill, and a lot of guys who were sufficiently exceptional to move forward despite their gender.

No, you’re misunderstanding. 

I’m saying that you want to hire the highest-skilled person, and you want to bring on-site everyone who has a reasonable chance of being the highest-skilled person, and one way to do diversity hiring, which I think is common in practice, is bringing someone onsite to see if they’re the highest-skilled person even if, based on their phone screen, we think there is only a 15% chance that they will in fact turn out to be the highest-skilled person. My entire argument is that this always involves hiring the most qualified person and instead of forcing you to trade diversity off against skill, lets you make a tradeoff between diversity and recruiting time.

I’ve had the chance to observe a lot of engineering hiring lately and it’s given me a lot of thoughts about diversity hiring.

Much of the screening process to hire for software jobs happens before you actually see an engineer write any code; you look at their resume and decide if they’re worth calling back, you talk to them on the phone and decide if they’re worth sending an at-home screen of some kind, you look at that and decide whether to bring them in for an interview. 

Now, obviously you can change around your standards for who you talk to and who you bring on for an on-site interview. There’s always a chance that a candidate will really impress you at the interview, even if they have a mediocre resume and aren’t impressive on a phone screen, but it’s not as likely. Maybe candidates who have pretty limited resumes and aren’t impressive on the phone have only a 5% chance of turning out, once you do an on-site, to be someone you want to hire. While candidates with a great resume who sound amazing on the phone have a 30% chance of turning out to be someone you want to hire.

Your engineering team does not want to interview twenty people to find someone hireable. They like writing software, not interviewing. If they sit through fifteen days of interviews with people who suck and aren’t hireable, they will be miserable. You will end up setting your bar higher, so that you don’t bother bringing people on for an onsite which they have only a 15% chance of passing, and you’ll inevitably pass up some great engineers you would have hired but you’ll also waste a lot less time.

So your engineering team wants more women/people of color/veterans/disabled people/etc. I feel like the narrative I’ve heard around affirmative action has been ‘either you can use it as a tiebreaker (i.e., when you have two equally qualified candidates you pick the candidate with the background you want) or you can pass up qualified candidates to pick less qualified ones or you can give up on diversity hiring altogether’.

And obviously there’s another option: you can be likelier to talk to them in the first place. You can let someone through to have an interview even if you think, based on the phone screen, there’s only a 15% chance that after the interview you’ll want to make them an offer. This does not involve hiring less qualified candidates, just talking to more people in order to miss fewer qualified ones. I think this is the most common actual diversity tradeoff involved in actual hiring, and it’s weird to me that it’s not even mentioned as a form diversity hiring can take when people are talking about the challenges of doing hiring appropriately.

(Incidentally the ‘tiebreaker’ thing is very very rarely what is going on, and pretty much never at a company like Google or Facebook. If Google interviews two great engineers they want, they’ll hire both of them, they don’t have to choose. They’re not trying to fill specific openings, they’re trying to get good engineers and place them on a team that has an opening.)

Anonymous asked: Why would you ever give an anecdote like that without linking to sources for it?

Sorry! See here, here and here for the whole story: John Shepperson learned that there were areas without power, took time off work, bought 19 generators out of pocket, and rented a UHaul to drive them down from Kentucky to an area of Mississippi without power after the storm. It would have been legal to sell them ‘at cost’ in the sense of ‘at their retail price’ but not ‘at cost’ counting expenses, let alone time and effort. He was jailed for it while the police confiscated and never distributed the generators. Note that the earliest version of the post had a serious error; he was only jailed for a week, which is horrible and unreasonable but much less bad than the version I first ran across.

There are two different things that both get called “price gouging”

fnord888:

They’re both characterized by a situation of sudden (and unpredicted) scarcity because of a breakdown in the usual supply chain that provides a good, and the price of that newly scarce good increasing dramatically.

One is where someone who already has a stock of the newly scarce good increases the price and reaps a windfall profit from the event. The other is where someone acts to increase the supply of the newly scarce good, and charges a price commensurate with the extraordinary measures required to do so (ordinary measures, by definition, no longer being adequate to provide a supply).

There are good reasons why we might want to treat these two cases differently, and yet I see very few people, on either side of the debate, willing to make the distinction.

In particular, if you put in tons of time and effort driving supplies to a disaster area from somewhere unaffected by the storm, you should be allowed to sell things for whatever price they sell at. Because getting more supplies to a disaster is good and if you’re not allowed to sell above a stupid definition of ‘at cost’ that doesn’t take into account ‘putting a thousand miles on my car’ or ‘losing my entire weekend’ or ‘the risk that I was wrong and this wouldn’t be needed’ then there will just be fewer supplies for disaster survivors.

And yes, the laws get used that way: After Katrina a guy heard that people needed generators, so he bought 19 of them in Kentucky, rented a U-Haul, and drove them to New Orleans. The police arrested him and confiscated the generators (which they did not distribute to disaster survivors). He intended to sell them at double the cost, and people were eager to buy them at that price. Instead he went to prison.

Anonymous asked: What's your point with the last post? The main reason Germany didn't have totally free speech is precisely to ban speech that aims at destabilizing the democratic state in order not to repeat what happened in the Weimarer Republik. So if you're post is supposed to criticize these rules, because they also apply to the left, I'm not sure what to make of that. Are you claiming the German justice system has a 'blind right eye'?? (Disclaimer: I'm German)

There are a bunch of American leftists who have been arguing recently that the United States should criminalize speech like much of Europe does, and in particular: 

1) that Germany bans speech by Nazis but refutes slippery slope arguments - the ban on Naziism has not resulted in the criminalization of speech they believe should be legal

2) that while people might think that a government with the power to criminalize speech will always end up using it for the benefit of the government rather than for marginalized people, in practice anti-free-speech laws do just get used against Nazis.

Shutting down a leftist website/forum because a post called to kill cops is a counterexample. It shows that European speech laws are not the narrowly-tailored anti-Nazi laws (which preserve the freedoms of anyone who isn’t a Nazi) they’re being sold as.

So I’m not accusing Germany of ignoring Nazis, I’m saying that the German laws are not just used against Nazis, and that we should not expect that if we had them here they would only be used against Nazis, and that ‘speech laws that stop Nazis without curtailing the freedom of anyone else’ are a pipe dream. Germany is not being hypocritical, they’re being consistent in application of a principle which I oppose.

Interior Ministry shuts down, raids left-wing German Indymedia site | Germany | DW | 25.08.2017

philippesaner:

disexplications:

Germany’s Interior Ministry on Friday banned and ordered raids on a portal popular with leftist readers and activists. Possibly the last posts from linksunten.indymedia.org - commemorations of a 1992 far-right mob attack on apartments where foreigners lived in Rostock-Lichtenberg and reports of racist graffiti on a memorial to a young woman killed by neo-Nazis in the United States - went live the previous night.

The site was closed for “sowing hate against different opinions and representatives of the country,” said Interior Minister Thomas de Maizière, adding that the operation of the site was now “a criminal offence.” …

Justifying the ban, de Maizière said that the measures were a “consistent” action against “left-wing extremist hate speech,” before adding, “The call for violence against police officers and their description as ‘pigs’ and ‘murderers’ is supposed to legitimize violence against policemen. It is the expression of an attitude that tramples on human dignity.”

European speech laws, everyone

Cops are never going to be more bothered by “kill jews” than by “kill cops”.

That’s one of the reasons that free speech protections are so important.

subbak asked: "Using some mechanism other than markets to decide what goods are produced in what quantities and how they are distributed: bad." Please tell me you don't believe that about services though (for example: it is good that railway companies keep operating lines that are losing money but necessary to some people).

I am totally in favor of this. I am in favor of it being done in a non-distortionary way, ideally by the government going “how much money are we willing to spend on transit access in rural areas? okay, let’s make that money available to whoever best provides transit access in rural areas”, rather than a distortionary/invisible costs way like “let’s mandate that anyone doing trains has to do some trains in rural areas”.

Anonymous asked: Do you see the emails of people supporting you on Patreon (in which case I should probably use that email?) or not?

I have access to it but if you say ‘I’m a Patreon supporter and want to request a thing’ I’m not going to go ‘the email doesn’t match, prove it’, I tend to assume good faith and this has always worked fine.

Anonymous asked: What is the best way to contact you about a Patreon thing? (I sent you a message on there but you said they're not working well)

Email. kelsey.piper2@gmail.com (And yeah, Patreon used to notify me when I got messages but now it does not and it makes them a hassle to check directly.)

Anonymous asked: Is communism good or bad?

  • Things that countries have tried to do with communism as a justification: bad,  universally so.

  • (Things that countries have tried to do with anti-communism as a justification: bad, universally so with the exception of the space program.)

  • Using some mechanism other than markets to decide what goods are produced in what quantities and how they are distributed: bad.

  • The Communist critique of the pain and dehumanization associated with low-wage work: basically correct.

  • the workers owning the means of production: neutral? like, this could happen in America right now, many workers in certain industries can afford to buy stock in the company such that they collectively owned a voting majority. They just generally seem to prefer to have their money in other forms which are more risk-averse and/or more liquid. Workers, when given the choice, do not seem to actually prefer to own the means of production.