The Silicon Uncanny Valley of Apps
Where is it creepy now?
Vicki’s note: I’m still figuring out my rhythm at work, and as such will be easing back into the newsletter. Expect upcoming posts to be shorter and the schedule to be kind of all over the place. Once I get back to what I feel like is a good cadence, I’ll reopen paid subscriptions.
Art: Pietro Longhi, The Letter, 1746
A few months ago, there was going to be an email revolution. Basecamp, the company behind the pre-Slack collaboration software, bought HEY.com and wrote that they were going to change email. Before, they wrote, email was a joy. But now,
[e]mail feels like a chore, rather than a joy. Something you fall behind on. Something you clear out, not cherish. Rather than delight in it, you deal with it. Your relationship with email changed, and you didn’t have a say.
Basecamp’s CEO, Jason Fried, and DHH, the CTO, then announced that Basecamp was working on a new app, a “love letter to email” that would be available in April. Given how high-profile both Jason and David are on The Internets, the app started to immediately generate buzz.
The fever pitch grew higher when Ryan, the CEO of Silicon Valley’s oft-watched product launch site Product Hunt gave a favorable review.
Finally, the app launched this summer, but the big news was not what the app actually did, but DHH’s fight against Tim Apple. Basecamp had a big argument with Apple, first in a tweet storm, then in the actual Apple Store. Eventually, Basecamp changed the app enough to appear in the Apple store anyway.
But none of that is even the interesting part of HEY, which I’ve been trying out on and off for the past couple days.
The most interesting part is that, in HEY, you’re the spam filter.
When you first sign in, it tells you about the Imbox (yes, they are doubling down on it.) The way you can land emails in your inbox is to have them go through manual approval the first time the app sees the sender.
What you do is approve senders to either continue send you email or not, and you do this for every single first-time sender, something that I’ve already done in Gmail through filters.
There are a lot of other rules about how HEY works, and they want to teach them to you over the course of 15 emails.
The most fascinating thing here is that HEY, as envisioned by its very opinionated anti-big tech founders, operates by very fundamental Normcore rules: no third-party data collection, thoughtful reflection, and a place that is ruled by human interests instead of the algo.
And boy do I dislike it.
Big G Wins Again
Maybe that isn’t a fair assessment. The HEY team has clearly thought a LOT about what email should be and worked very hard to execute on it. What I mean is that, after operating under the thumb of Big Data Google for so long, ever since they released Gmail as an April Fool’s Joke in 2004 actually, I am used to the way Gmail works and I’ve adapted to its workflows for my purposes.
I hate that I am used to it since it constantly indexes my email in order to send me ads and does who knows what else with the data, but I’ve already set up hundreds of manual filters for mail I don’t want. The spam blocking just happens naturally on its own. Every email comes into my inbox and I deal with it as it happens, or it stays in my inbox, and when I answer it, I archive it. I don’t want to set it aside into a separate queue. I don’t want a separate place for my receipts or phone numbers - I can use the Gmail search box just fine.
I get that these are all pain points and that the people working on HEY have worked very hard to try to address them. But there is something off about the way they’ve de-machine learned email, and I think that the answer is that when we know The Black Box is doing the work for us, we kind of throw our hands up and shrug. “It’s a machine, so it’s stupid and it’s a blunt tool, but it works for most of our use cases.”
The Uncanny Silicon Valley
But when we know that it’s mostly people with opinions about how UX should work behind the scenes, we are suddenly suspicious. Because every human has their own opinion on how things should work, and if they’re tweaked to one person’s workflow, all of a sudden they don’t work for other people. Take for example every time sites, like Twitter, have site redesigns and users complain. The site redesigns are based on UX studies, sure, but they’re still based on a very specific subset of user needs and are more or less manually changed by UX teams.
And the closer we get to the other side of this uncanny valley where there are people making decisions, the angrier people get as well, because we get to the “Why Wasn’t I Consulted” point. If you know there is a designer or a DHH on the other side, it’s easy to blame them for their shortsightedness. And in fact, many have.
(Later in the thread, DHH says he’ll work to address this issue.)
So really, the sweet spot for apps is to have some sort of algorithmic input, but like Michael Pollan says, not too much.
It looks something like this:
When there is a person completely controlling the app, you can feel like you’re being watched, and when there is an algorithm completely controlling the app, it feels also unsettling in that no one is helping you and you’re at the mercy of The System. It’s only when the two work together that things work well.
The interesting thing is that there are a lot of companies, having left everything to chance and algorithms, will move in the human direction in the next five years. We’ve spent enormous amounts of time getting to the point where we think machine learning dictates things, only to find out that it falls apart and requires manual intervention.
Here are some recent examples of cases where people at various companies have decided to intervene to counteract the prevailing algorithms:
Twitter manually hiding tweets
FB refusing to and then also taking down violent posts
Pinterest deciding to disallow anti-vax content (by the way, you should read this Twitter thread by the community manager leading that effort and how terrible it was for her at Pinterest)
YouTube banning a number of vloggers
HBO curation is going manual to combat Netflix’s algorithms
YouTube banning a chess blogger for casually mentioning COVID and unbanning him manually because he’s popular enough that people complained
This is just a small smattering. Sure, a few points don’t indicate a trend and I’m not saying we’re all shutting down the Spark clusters chugging away on recommendations anytime soon. But I think what we’re finding is that figuring out which side of the curve to be on on this stuff is very, very hard, and the big companies are at the point where, ironically, they’re moving back to the point where people start to matter again and starting to face some competition in that space.
Or so we hope.
What I’m reading lately
Just finished Skunk Works, which I now recommend to everyone in spite of its antiquated views on diversity in the workplace. Important read about both project management and the history of the American military/space programThis story on opioids is just nuts
Tracy’s post on being a woman in tech is Very Good
A day in the life of two different Indian students under lockdownWhat do you hate about dashboards?
How product work happens
This newsletter’s M.O. is takes on tech news that are rooted in humanism, nuance, context, rationality, and a little fun. It goes out once a week to free subscribers, and once more to paid subscribers. If you like it, forward it to friends and tell them to subscribe!
I’m a machine learning engineer. Most of my free time is spent wrangling a preschooler and a baby, reading, and writing bad tweets. Find out more here or follow me on Twitter.