Worse than that. Because liking One Direction was embarrassing, most of them never revealed to anyone in their real life that they did this online.

I decided I had to find out more.

So, I became a Larry.

I created an online identity, and made my own tumblr.

I reblogged cute photos of Harry and Louis, and made friends.  

Great friends, actually. Shoutout to my groupchat Larries.

It took a couple of weeks. You can’t just turn up on tumblr and ask a question. It’s like being an egg on twitter and hoping someone sees what you have to say.  But eventually, I had a tiny number of followers who might have been prepared to share something that I’d made, and so I threw together a survey.  

People who do research for a living will probably weep about my scientific method. I’ve had nearly 600 responses, and if I’d had any idea the response rate would be that high I would have agonised more over the questions, but hopefully you get the idea.

First, I asked them what platforms they used to participate in the fandom.  As you can see, asking the question on tumblr skews the demographics.  The second most popular response there is AO3, or Archive of Our Own, where fanfiction is hosted. I’ll talk more about that in a second.

Then, I asked them what works they’d created or contributed to.  Again, you can see writing is the most popular but there are a range of more technical contributions.

I asked them if they had ever considered a career in technology.  Overwhelmingly, they said no.

Then the important question. I gave them a free text box, and I asked them “If not, why not?”  The answers are heartbreaking.

“I'm tech savvy but self taught. haven't taken classes”

“I do not have the education.”

“Despite the modern world I've grown up in it still seems as though I wasn't made aware that tech is a field of work and study. I couldn't name a job position/career or course that would be considered 'tech'.”

“I don't think I'm good enough to actually get far enough ahead that I would be hired over someone else”

“I'm intimidated by technology”

“When I was at a career-picking age tech was not for girls. I taught myself some html, but there really weren't opportunities for women who were less than 100% certain and willing to fight for it. “

“I just don't think I'd be good at it. I can make videos that I enjoy but in the real world for a job I can't make a dumb little four minute Larry music video haha”

“I don't feel skilled enough to enter the industry”

“I don't think i'm talented enough”

Bear in mind - 70% of these respondents are aged 15-24.  All but five of the respondents identified as female or non-binary. 65% identified as a sexuality other than straight (recognising that has something to do with tumblr as a platform).  That’s our diversity pipeline. Right there.  Glued to their computers. And not telling anyone about it.

It also turns out that they’re wrong:

We’ve thought for a long time that our young women underperform young men in STEM subjects, particularly at high school.

In 2014, the National Assessment of Educational Progress (NAEP) administered the first-ever nationally representative assessment of technology and engineering literacy. Eighth-grade students were presented real-world scenarios involving technology and engineering challenges. Students were asked to respond to questions aimed at assessing their knowledge and skill in understanding technological principles, solving technology and engineering-related problems, and using technology to communicate and collaborate.  Eighth-grade female students scored higher on average than their male peers in TEL overall. 

We also know now that diverse teams are critical to building successful products.  

The New Yorker did a great profile of a hackathon held by the developers for AltSchool, an education startup in the US that makes significant use of technology. The hackathon was designed to come up with new solutions for teachers in the classroom.

One of the teams had focussed on bookmarking video. The team had decided to try to find a “fun route” to help teachers request a video clip of a moment in class. “The idea is that the teacher could, in theory, just knock twice on their phone,” one team member said. He patted twice on his device, which was buried in the front pocket of his jeans, to demonstrate the ease and unobtrusiveness of the gesture.

From the back of the room, a woman spoke up: “Did you test it with a woman?”

Many participants laughed.

"I’m serious,” the questioner went on. “A lot of our teachers are female, and they carry phones in different places.”

The members of the bookmark team, all of whom were male, looked deflated. In coming up with their apparently elegant solution, they had not visualized a female teacher slapping her bottom to activate a phone tucked into her back pocket.

Google's speech recognition software, which is currently being used to autocaption YouTube videos, is capable of correctly captioning a female voice less than half that time, but can correctly caption a male voice over 60% of the time. 

We all laughed when the Washington Post made this epic blunder on their cover in relation to the Women's March.

But how many of us stopped to ask why Wikipedia defaults to the male symbol even when you're searching for the female one?

We see it all the time in software, when teams aren’t diverse.

We even see it in hardware, VR headsets that make women nauseous, or touchscreens that don’t take account of women having a lower core temperature.  Women buy or influence 85% of all consumer purchases, while 85% of product designers and engineers are men. 

And that’s just talking about gender.
When Pokemon Go swept the world, it didn’t take long for some players to discover that the pokestops and gyms were concentrated in predominantly white neighbourhoods.  Thinktank Urban Institute in the US found that the difference was stark.   Looking at data for Washington DC, they found that in neighbourhoods that are majority white, there are 55 portals on average, compared with 19 portals in neighbourhoods that are majority black.

Why? Because Pokemon Go’s dataset came from the earlier game Ingress.  Ingress players tended to be younger, white, English-speaking men - tech early-adopters, and because Ingress’s portal criteria biased business districts and tourist areas, portals ended up in white-majority neighbourhoods.

But in launching Pokemon Go off this dataset, no one noticed, or cared enough to address the bias.

At home in New Zealand, our own passport photo recognition software can't recognise Asian faces.

All the studies show that companies with racial, gender and ethnic diversity do better overall. 

And yet we’re still struggling to change it, and one of the things we often blame is the “pipeline”.  People who are hiring don’t want to “lower the bar”, and claim there just aren’t enough qualified, diverse candidates out there. 

So here’s my challenge.

How do we get these young women to see themselves as skilled and talented enough to work in our industry, and how do we change the way we think about hiring and qualifications so that there’s a path for them to join us?

We’ve already seen fan writers enjoy runaway success as “real” authors.  

What’s the dev equivalent of that?

Before, I was talking about fan works uploaded to Archive of Our Own.  It’s worth chatting about that for a second.

The Archive of Our Own (AO3) is a noncommercial and nonprofit central hosting site for transformative fanworks such as fanfiction, fanart, fan videos and podfic. The Archive entered open beta in November 2009. In December 2015 the Archive passed two million uploaded fanworks, and in 2016 it is used by over 900,000 registered users, and has fanworks in over 22,000 fandoms.

The Archive was entirely built and designed by volunteers from fandom. Many of the volunteers acquired skills in coding, design and documentation through their work on the project. When the project started, they asked volunteers who had never coded before to read an intro to both Python and Ruby, and then to write a very short programme. They asked the volunteers what their experience was like. They chose Ruby as the language based on that.  When the project went in to open beta, it had over 20 contributors, all of whom were female, and it remains a shining example of an open source project that has created an inclusive environment that welcomes people without “traditional” experience.  

In a similar vein, this is my friend Maciej. He loves this conference, and he’s the one who suggested I come here. You can see copies of his talks at his website idlewords. This is him in Havana a couple of months ago, and it was a face-melting 34 degrees, but if you follow him on twitter you know that he refuses to wear shorts. So here he is, toughing it out in jeans.

One of the talks he gave a few years back is called “Fan is Tool Using Animal”, and it was about when Delicious, the bookmarking site, made a bunch of changes that made the site unusable for the fans who had been using it to meticulously tag their fan fiction collections.

Maciej also runs a bookmarking site, and so he tweeted fans to ask what they might need.  You should go read his talk, but basically over the next couple of days, fans came together anonymously to collaborate on a google doc, and produced for Maciej a 52 page tech spec.

“Having worked at large tech companies, where getting a spec written requires shedding tears of blood in a room full of people whose only goal seems to be to thwart you, and waiting weeks for them to finish, I could not believe what I was seeing. It was like a mirror world to YouTube comments, where several dozen anonymous people had come together in love and harmony to write a complex, logically coherent document, based on a single tweet.”

Many fans don’t value their skills, in part, because they’re self-taught.  But let me ask you a quick question.  How many people in this room, whether you’re designers, or developers, how many of you started out on a self-taught pathway with computers?

My self taught pathway started with this baby. My first computer was a Commodore VIC 20

I taught myself BASIC out the back of a magazine when I was ten years old. 

I taught myself HTML to build my first website several years later. 
(Check out those sweet tables)

Know what it was for? A place for my fanfiction. Know how many people in my “real life” knew that I’d done it? None. I've never talked about this publicly until today.

I’d been socialised out of using computers at high school, because there weren’t any girls in the computer classes, and it wasn’t cool, and I just wanted to fit in.  I wound up becoming a lawyer, and spending the better part of twenty years masquerading as someone who wasn’t part of the “tech” industry, even though basically all of my time was spent online.

And I can’t begin to tell you how common it is. 

So what if your first experience of “code” is cutting and pasting something to bring back replies because Tumblr took them away and broke your experience of the site.

Is that any more or less valid than any dev cutting and pasting from Stack Exchange all day long?

What if your first online experiences were places like Myspace and Geocities. Or if you started working with Wordpress and then eventually moved into more complex themes and then eventually into plugin development? Is that more or less valid than the standard “hacker archetype”? Aurynn gave a great talk recently about the language we use to describe roles in tech. How “wizards” became “rockstars” and “ninjas”.  But also, and crucially, how we make people who haven’t followed a traditional path feel excluded.  Because they haven’t learnt the “right” programming language, or they haven’t been programming since they were four, or because, god forbid, they use the wrong text editor.

There might be really good reasons why super talented candidates haven’t been programming since they were four.

So when we’re hiring for roles, for junior graphic designers, and junior front end developers. Photo and production editors. For social media managers. 

What do we need to change?

Our first problem is pattern matching: When we look for traits in candidates that we think will point to successful hires, because we’ve seen those patterns work before.

But it's like trying to put together the Avengers, and only hiring Hawkeyes.

Pattern matching is a problem found while reviewing resumes, from universities to previous companies worked. And again when we’re interviewing.  It’s not so much just that we hire people who actually look like us. But that we hire people who have been on a similar journey to us. Who are interested in the same things we are.  Who pass our invisible “culture” tests, like would we want to grab a beer with them? Or sit on a plane for 12 hours with them?

It’s what Mitch Kapor calls the “mirrortocracy”. We think we’re hiring for merit, but we’re hiring mirror images of ourselves.

We’ve started throwing around the terms “unconscious bias” or “implicit bias” - and for some people they’ve started to be seen as codewords that mean they’re just being called out for being sexist or racist. 

But it’s actually a cognitive tool that helps us get through every day.  It’s all the tiny assumptions that we make as we navigate the world because we’re bombarded with too much information and our brains sort it quicker than even we can recognise. And if we keep thinking of unconscious bias as an insult - or as something we can fix or eliminate by going to a single workshop - we’re missing the point.

Project Include is a group effort to make specific recommendations to tech companies around improving diversity and inclusion, not just gender and not just in terms of hiring but company culture, remuneration and so on.  It’s a great site, with a lot of amazing resources. Including for small teams.

There are a couple of specific recommendations that are useful in this context. One of these is to try and eliminate bias in resume evaluation.

They link to studies that have shown systemic unintended bias when reviewing resumes that are identical with the exception of names that signify a racial background or gender or a resume entry that signifies LGBT status.

Recently, venture capitalist John Greathouse got in a pile of hot water for this tone deaf op ed he wrote for the Wall Street Journal, in which he acknowledged this unconscious bias, but suggested the answer was for women to hide their gender to get ahead. It was met with predictable ridicule, and to be fair, he apologised for it unreservedly, calling the article dreadful. 
There are actually a range of ways we can use technology to remove signifiers off resumes before reviewing them, and services that will do that for you.

But I think the more interesting question is what would it look like to really hire blind? Not just to take names and genders and ages off  CVs, but to do away with resumes altogether and hire people who can do the job.

There’s a startup that’s come out of Y Combinator called TripleByte that’s trying this with developers on the basis that most programmer interviews are broken.  As they say, credentials should not be used as a proxy for talent. Education and work history are meaningful but relying solely on them results in missing good programmers. Good programmers come from all types of background. It's what you can do that matters, not where you went to school. 

TripleByte vets the programmers themselves, and then recommends them to the companies hiring. Importantly, they’re not interested in your cv.  For the companies that use TripleByte to find them candidates, they trust that they’re going to be recommended people who can do the job.

GapJumpers enables companies to host “blind auditions”.  Their data shows that women are taking blind auditions at a rate comparable to their representation in the US general population - around 54%.  Blind auditions are also taken by applicants from diverse educational backgrounds. i.e. state schools and universities, community colleges, MOOCs, coding bootcamps, and self taught candidates. Women make up almost 60% of top performers in their blind auditions. So, what GapJumpers sees is that by interrupting implicit bias in real-time, blind auditions allow employers to hire from a skilled and diverse slate of talent.

More than that though - we need to actively look for candidates in non-traditional spaces.  Project Include notes that the location of job advertisements can make a big difference. If ads are placed in locations frequented by the industry and conventional networking groups, diverse applicants may not find them. We need to think about local open spaces, like coffee shops, community centres, and libraries. Online posts in different communities. Building a network of people who will recommend different candidates to us.

And if we want to attract a different kind of candidate, we need to think hard about our job descriptions.  

The point is not to hire diverse candidates because they’re diverse. The point is always to hire amazing candidates. But if only men are responding to your ad, you have to ask yourself? Was there something about the ad that only men were interested in? Did I put the ad somewhere only men would see it?

What if instead of saying “7 years plus experience” you tried “experience solving these kinds of problems” and listed them?

Do you really care if the person has a compsci degree? Or is that just on your wishlist? Because if you put it in the ad, you’re turning a bunch of people away.

Textio uses machine learning to compare your job post to more than 40 million others to predict how well it will perform before it’s ever published.  A high score on their platform attracts applicants that are more qualified, more diverse, and the positions are filled faster.

The team at Buffer changed just one word.  
Developers at Buffer all used to be called “hackers.” They had front-end hackers, back-end hackers, Android hackers, iOS hackers, traction hackers.
They started using the word “hacker”in Buffer’s early days because it felt like the most inclusive way to describe the work developers were doing. The CTO said:
“I think the original reason why we liked that word was because hackers are just people who get things to work well and fast. A ‘hacker’ doesn’t necessarily need a computer science degree or a lot of experience or need to be excellent in mind games, puzzles etc.”

But as they began to grow, Buffer was seeing a very low percentage of female candidates for developer jobs—less than 2% of candidates.  When they dug in to  this challenge, they discovered the “hacker” title wasn’t as inclusive as other titles and could be tough for many to identify with. 

They opened up the discussion inside the company, and tried to brainstorm what better words they could use: Engineer, Developer, Maker, Builder, Programmer, Specialist, Artisan, Architect, Helper, Code experimenter
In the end they went with Developer, but they also decided there was a lot more work for them to do.  One of the cool things they now have up on their site is a Diversity Dashboard, so candidates can dig in and see the kind of company they might be joining.

LinkedIn has gone a step further, offering a six month full-time paid internship for software developers from non-traditional backgrounds. They encourage applications from candidates without computer science degrees who are self-taught, re-entering the workforce, starting second careers, veterans, or those who have attended bootcamp-style programming courses.

If there’s one thing I know for sure it’s that diversity doesn’t have a single solution. We need to work on culture and unconscious bias, we need to make our computer science degree courses practical and interesting and welcoming so that more people want to take them, we need to improve technical education in our schools so that our kids are as excited about a possible career in tech as they are anything else. We need to improve access to technology in our communities in the first place.

But when it comes to just this one aspect of the problem, our pipeline, maybe we also need to recognise that that it’s already full of talented, creative, passionate, dedicated, hardworking young women online.

We just have to find a way to include them.

With thanks to fan artists: @yoursincerelylarry @ablyablya @larryswand @stylinsinz @fxckingunicorn @claudiyah @firsova @fayestardust @alistruecolors @ifhalocouldfly @stylinsonarchive

And clique Larries: @allegedlymags @scrufflecake @allthelivelonglarry @ellemem @niallgoestothezooalone @jewleeahhhh @sammie4jones

And to @ksuyin for the slide deck