Pixel Envy

Written by Nick Heer.

Archive for August, 2014


Anand Lal Shimpi:

I’m 32 now. The only things that’ve been more of a constant in my life than AnandTech are my parents. I’ve spent over half of my life learning about, testing, analyzing and covering technology. And I have to say, I’ve enjoyed every minute of it.

But after 17.5 years of digging, testing, analyzing and writing about the most interesting stuff in tech, it’s time for a change. This will be the last thing I write on AnandTech as I am officially retiring from the tech publishing world.

I wish Lal Shimpi the very best, but I’m going to miss his fastidious, detailed, unique take on the tech world.

“Wipe My Ass With Your Feelings”

Nilay Patel is grouchy:

Today, my friends at Vox.com published a terrific 5,000-word feature about the legacy of the Sopranos, framed around one very exclusive piece of reporting: series creator David Chase told reporter Martha Nochimson whether Tony Soprano dies at the end of the show, a question that fans have debated endlessly in the decade since the series famously ended on a hard cut to black.

It’s terrific, and the Vox.com product team engineered a fantastic presentation where the screen blacks out before the reveal. It’s everything a feature on the internet should be: thoughtful, concise, exclusive, and interactive.

But because the headline was phrased in the form of a question — the question of the entire series — Jake Beckman, who runs the Twitter account @savedyouaclick, decided that it wasn’t worth it. He “saved you a click” and tweeted the reveal.

This is bullshit.

No it isn’t.

If the article is so dependent on the teaser headline that a single tweet can bust the whole thing up, then Beckman did save people a click. If the article is not dependent on the teaser headline and it can stand on its own, why use that particular headline? It attracts clicks, but at the cost of feeling a little trashy.

Put another way, imagine if the Wall Street Journal redesigned their paper to look like the National Enquirer. Would you find it as trustworthy?

Facebook to Reduce Click-Bait Headlines With This One Weird Trick

Khalid El-Arini and Joyce Tang of Facebook:

So how do we determine what looks like click-bait?

One way is to look at how long people spend reading an article away from Facebook. If people click on an article and spend time reading it, it suggests they clicked through to something valuable. If they click through to a link and then come straight back to Facebook, it suggests that they didn’t find something that they wanted. With this update we will start taking into account whether people tend to spend time away from Facebook after clicking a link, or whether they tend to come straight back to News Feed when we rank stories with links in them.

Facebook’s tracking code makes this sort of thing possible, which seems both more useful to me than serving me ads, and edging closer to the creepy line. But this serves to reiterate the point Zeynep Tufekci made in her article about the algorithms used by Twitter and Facebook to highlight stuff we might be interested in: what makes this code ideal to decide what is important to us? There’s a lot of responsibility inherent to a deployment of an algorithm update used for user feed selection and sorting, but it increasingly feels as though it’s not being given the respect it deserves by Facebook and Twitter staff.

Samsung’s Ice Bucket Challenge

This is a pretty slick ad: take a very current “viral” thingy and spin it in a short time to point out a great advantage Samsung’s product has over its competitors’. And they apparently gave a very generous donation to charity, so it’s good all around, right?

Gruber notes two things, though: it’s more like a promotion of the Galaxy S5 rather than truly dedicating itself to the charitable cause, and the status bar changes midway through the ad. Regarding the former point, I don’t see how this is much different than the Product Red campaign and its associated products. Regarding the latter, I would hope that this is due to a filming error or something. But knowing how unscrupulous Samsung has been in the past with regards to their ads, I wouldn’t be surprised if this whole thing were faked.

PS: If you want a better ice bucket challenge video, Jeremy Clarkson’s kids have you covered (NSFW language, obviously).

Your Twitter Timeline Is More Than Just Tweets From People You Follow

Dan Frommer, Quartz:

In addition to the basic, essential definition of a Twitter timeline — “all Tweets from those you have chosen to follow on Twitter” — plus retweets and ads, there’s a new section:

Additionally, when we identify a Tweet, an account to follow, or other content that’s popular or relevant, we may add it to your timeline. This means you will sometimes see Tweets from accounts you don’t follow. We select each Tweet using a variety of signals, including how popular it is and how people in your network are interacting with it. Our goal is to make your home timeline even more relevant and interesting.

In most cases, these seem to be tweets favorited, but not retweeted, by people you follow.

The official Twitter apps and website are becoming increasingly worse; it’s no wonder users are flocking to third-party apps.

Ferguson and Police Militarization

Speaking of Ferguson, here’s John Oliver’s thorough, thought-provoking and — occasionally and blessedly — hilarious take on the shameful reaction by the police department and public officials so far.

What Happens to #Ferguson Affects Ferguson

Zeynep Tufekci:

Maybe, just maybe, there can be a national conversation on these topics long-ignored outside these communities. That’s not everything: it may be a first step, or it may get drowned out.

But at least, we are here.

But I’m not quite sure that without the neutral side of the Internet—the livestreams whose “packets” were fast as commercial, corporate and moneyed speech that travels on our networks, Twitter feeds which are not determined by an opaque corporate algorithms but my own choices,—we’d be having this conversation.

There are already a lot of conversations surrounding the tragic and complex problems in Ferguson, but plenty more need to happen. The White House has pledged to investigate the militarization of local police, for example. But it’s worth asking how many of these conversations would be happening had the algorithms that inform our conversations selected otherwise.

California Über Alles

Peter Christiansen commenting on Hacker News on my little ditty about diversity:

I wish they would have included demographics for the Bay Area, not just the USA. From Wikipedia – 52.5% White, 6.7% non-Hispanic African American, 23.3% Asian, 10.8% from other races, 5.4% from two or more races, 23.5% Hispanic or Latin. (Incidentally, this almost exactly matches Apple’s numbers, except for Hispanic). Comparing tech worker demographics to Bay Area demographics, It’s kind of a chicken vs egg about whether the demographics precede the jobs or vice versa, but given how geographically concentrated the tech industry is, this seems like a bad oversight.

This was by far the most common critique I received about the piece: why did I compare tech company demographics against the US as a whole, and not just California?

While it’s true that the American West Coast has typically had a much higher Asian population than, say, Cleveland or Mobile, it’s hard to say whether it has grown disproportionate to the rest of the United States, as records have been pretty poorly kept. This is what Christiansen alludes to in his comment: is the tech industry responsible for the demographics of the Bay Area, or are Bay Area demographics fairly represented by most of these companies?

Apple on Diversity

Khoi Vinh on Apple’s (and others’) employee diversity figures:

[M]ost retail employees are likely part-time and/or relatively low wage earners; what would these numbers look like if they were segmented so that we can see how well Apple’s diversity initiatives are faring for full-time workers earning over $100,000 a year? Or full-time workers earning more than $200,000 a year? I suspect the numbers would then look less encouraging, maybe even starkly different from what’s being reported here.

Vinh makes a good point — employees in “leadership” positions are universally more white than employees in tech or non-tech; similarly, they’re also typically more male than non-tech (though not more male than tech workers). The stereotype is that Silicon Valley is run by white men; these figures are the proof.

Diversity of Various Tech Companies By the Numbers

See Also: The 2015 edition of these numbers, adding Amazon to the mix.

With Apple’s report today (finally), major tech companies have all published information about racial and gender diversity. I thought it might be useful to run the numbers and compare them against the demographics of the United States as a whole, for reference. All data is as-reported from each company.

Gender Diversity

Almost all available data in this selection of companies solely reports a male/female split. Yahoo is the only company that has an “other/not disclosed” option.

Update: Apple, Google, and Microsoft all have retail operations which are not made distinct from the corporate side. (Thanks to Krishnan Viswanathan for pointing this out.)

Gender Diversity, USA
Category Male Female
USA Overall (approx.) 49% 51%
USA Workforce (PDF) 53.1% 46.9%
Gender Diversity in Tech Positions

Microsoft does not separate tech and non-tech workers, so their data in this table is the same as their data in the next one.

Company Male Female
Apple 80% 20%
Facebook 85% 15%
Google 83% 17%
LinkedIn 83% 17%
Microsoft (see note) 76% 24%
Twitter 90% 10%
Yahoo (see note) 85% 15%
Gender Diversity in Non-Tech Positions

Microsoft does not separate tech and non-tech workers, so their data in this table is the same as their data in the next one.

Company Male Female
Apple 65% 35%
Facebook 53% 47%
Google 52% 48%
LinkedIn 53% 47%
Microsoft (see note) 76% 24%
Twitter 50% 50%
Yahoo (see note) 47% 52%
Gender Diversity in Leadership/Executive Positions

The “USA” row uses the “management occupations” data from the BLS document above, as a rough and imperfect approximation.

Microsoft’s data is too poor to use for this table, as they lump ethnic minorities and women into the same statistic.

Company Male Female
USA (PDF, pgs. 21-23) 62% 38%
Apple 72% 28%
Facebook 77% 23%
Google 79% 21%
LinkedIn 75% 25%
Microsoft (see note) N/A N/A
Twitter 79% 21%
Yahoo (see note) 77% 23%

Ethnic Diversity

The “USA Workforce” row uses data provided by the Bureau of Labor and Statistics (PDF). Their demographics information (indicated page 9) is kind of a pain in the ass, though: the unemployed column is a percentage of the labour force, but the employed column is a percentage of the total population. I’ve done the math, though, and the results are what’s shown below. In addition, the BLS does not separate out those of Hispanic descent because “[p]eople whose ethnicity is identified as Hispanic or Latino may be of any race.” As such, the row will not add to 100%, but the percentage of Hispanics in the workforce has been noted per the table on page 10.

Similarly, the “USA Overall” row uses data from the CIA World Factbook, and they, too, do not note those of Hispanic descent separately. This row will also not add to 100%.

Apple, Google, and Microsoft all have retail operations which are not made distinct from the corporate side.

Ethic Diversity, USA
Category White Asian Hispanic Black Mixed Other or
USA Overall 79.96% 4.43% 15.1% 12.8% 1.61% 1.15%
USA Workforce (PDF) 80.5% 5.4% 15.3% 11.1% 1.6% 1.2%
Ethnic Diversity in Tech Positions

Microsoft does not provide a thorough breakdown of their racial diversity data; their data in my table is lumped into the “Other” category. They also do not separate tech and non-tech workers, so their data in this table is the same as their data in the next one.

Company White Asian Hispanic Black Mixed Other or
Apple 54% 23% 7% 6% 2% 8%
Facebook 53% 41% 3% 1% 2% 0%
Google 60% 34% 2% 1% 3% <1%
LinkedIn 34% 60% 3% 1% 1% <1%
Microsoft (see note) 61.8% N/A N/A N/A N/A 38.2%
Twitter 58% 34% 3% 1% 2% 2%
Yahoo 35% 57% 3% 1% 1% 2%
Ethnic Diversity in Non-Tech Positions

Microsoft does not provide a thorough breakdown of their racial diversity data; their data in my table is lumped into the “Other” category. They also do not separate tech and non-tech workers, so their data in this table is the same as their data in the previous one.

Company White Asian Hispanic Black Mixed Other or
Apple 56% 14% 9% 9% 3% 9%
Facebook 63% 24% 6% 2% 4% 1%
Google 65% 23% 4% 3% 5% <1%
LinkedIn 63% 26% 5% 3% 3% <1%
Microsoft (see note) 61.8% N/A N/A N/A N/A 38.2%
Twitter 60% 23% 3% 4% 5% 5%
Yahoo 63% 24% 6% 3% 2% 2%
Ethnic Diversity in Leadership/Executive Positions

The “USA” row uses the “management occupations” data from the BLS document above, as a rough and imperfect approximation of the broad US national trend.

Microsoft’s data is too poor to use for this table, as they lump ethnic minorities and women into the same statistic.

Company White Asian Hispanic Black Mixed Other or
USA (PDF, pg. 24) 85.7% 5.1% 8.8% 6.9% N/A N/A
Apple 64% 21% 6% 3% N/A 6%
Facebook 74% 19% 4% 2% 1% 0%
Google 72% 23% 1% 2% 1.5% <1%
LinkedIn 65% 28% 4% 1% 3% <1%
Microsoft (see note) N/A N/A N/A N/A N/A N/A
Twitter 72% 24% 0% 2% 0% 2%
Yahoo 63% 24% 6% 3% 2% 2%


Let’s get something out of the way: I’m a white twenty-something Canadian who graduated from art college. Analysis of statistics of racial and gender diversity at American tech companies is not exactly my strongest suit. But, hey, you’ve made it this far. I want to be as fair as possible to everyone represented in these stats. If there’s a problem, please let me know.

  • As noted above, the data available from all companies only reports a male/female split. While it would be imprudent for an employer to ask for more information, it does misrepresent individuals of other genders.
  • It will come as no surprise that all of these companies are boys’ clubs, particularly tech workers and those in leadership roles. This is one of the biggest issues facing the tech industry right now.
  • Stereotypes are proving quite strong with the significant over-representation of those of Asian descent at all companies surveyed.
  • Black employees are, on the other hand, significantly under-represented. Like the under-representation of women in tech companies, this suggests a much larger and more overreaching issue. I’d argue that this is another of the biggest issues facing the tech industry.
  • Only a single data point was typically made available in a given category. Microsoft was an exception, showing how their diversity has changed over the past few years. I think it would be valuable for the surveyed companies to release similar data from past years. I mention this not because I want a feel-good kind of statistic, but because I’d like to see if progress is, indeed, being made, and at what rate.
  • Generally, only ethnicity and gender data was provided by the companies surveyed. As several of the reports stated, diversity is so much more than just these two genetic features. It would be inappropriate for employers to ask about sexual orientation, childhood household income, and so forth, but these qualities are part of what shapes internal diversity. Poor families — or even most middle-class families — can’t afford to send their kids to Stanford.

Opinionated Software

Guy English:

[B]eing opinionated isn’t the goal. Being useful is.

Being opinionated and shipping the truest form of your vision of software doesn’t assure success. I understand the amount of heart, soul, concentration and perseverance it takes to ship a piece of software that really makes you proud and hits all of the marks you’d set for yourself and your team. It can be a really great piece of software.

That doesn’t mean it deserves to be a hit.

I agree with English, but developers who do release “opinionated” software have got to be aware that the more specific they make their software, the smaller the market gets. If you’re a developer and you’re putting out a sweet new app that’s “opinionated” and you don’t recognize that you’re therefore limiting your market, you’re lying to yourself.

What It Was Like to Do Surprise Improv With Robin Williams

Chris Gethard, writing for Vulture:

But this show also stresses me out. Because I now organize it, I put a lot of pressure on myself to put together the best possible casts. So I am always bothering the improvisers who are on Saturday Night Live, the writers from Colbert Report, and the guys from Conan’s staff to come do it. And they’re busy people, and I hate being the one that bothers them each week with my dumb text messages begging them to come do this show, because not only am I being very annoying, it’s also a weekly reminder that I’m not quite where I want to be. I would like to be the person who gets annoyed, not the person who does the annoying.

And on this night, I’m in the back of the theater, tired and stressed out by all this self-defeating thought, exhausted by it before the show even begins.

And Robin Williams walks into the green room.

Robin Williams

There are probably half a dozen things that I’m thinking of posting today, but this rose right to the top of the queue. It isn’t often that the death of a public figure really gets to me, but this is one of those occasions. It’s like losing your best friend; Robin Williams often felt like everyone’s best friend. It’s a tragic way for a brilliant mind to go.

Ten Years of the Impulsive Buy

Impulsive Buy editor-in-chief Marvo who, like Madonna, has no apparent surname, is celebrating ten years of reviewing mostly junk food:

Here are other numbers that might interest you: 3,306 posts, 1,671 reviews, and 30 pounds gained over the past 10 years. Man, if only I took pictures of my belly every day for the past 10 years, it would’ve made a great YouTube video. And when I say “great,” I mean “gross.”

The Impulsive Buy is one of my very favourite regular reads. Congratulations to Marvo for ten years of showering clothed and brushing your teeth with wasabi in the name of making the best grocery reviews on the web.