I’ve been experimenting with a daily email with Colin Nagy called Why Is This Interesting? This is from today’s edition. If you’re interested in checking it out, drop me a line (I’ll post something here when we launch in publicly).
If you’ve spent any time working in the age of e-mail (nevermind Slack) you’ve encountered this challenge. One of the things I shared with everyone who started at Percolate for a long time was this post fromY-Combinator founder Paul Graham about the schedule a “maker” keeps vs a manager. The point is that the manager has their days broken into tiny bits, 30 minute or one hour meetings, while the maker needs long uninterrupted focus time to do their work. When the manager forces the maker into their schedule they are surprised that the work can’t get done.
One way to think about this divide is as something computer scientists callexploration vs exploitation. The manager is an explorer, looking at information across many different areas, while the maker is an exploiter, using that information to go deep in just one. It’s a little like the story from the Greek poet Archilochus about the fox and the hedgehog: “the fox knows many things, but the hedgehog knows one big thing” (if you’re not familiar with the parable,here’s a good primer from NPR).
As Brian Christian points out in his bookAlgorithms to Live By: The Computer Science of Human Decisions, there’s actually tons of interesting stuff that lives in this tension. Do you go to the restaurant you like or try the new one that just opened? Should you load up an old favorite on Spotify or see what they’ve chosen for you this week? The answer, as we have all figured out and computer science has proven, is it depends. To figure out the best approach you’ve also got to know the time limit. In simplified terms, if you have lots of time left, exploration makes sense, if you’re approaching deadline, exploitation is optimal.
I started and stopped this post four times as I tried to find the right way to open. Eventually I got tired of searching and figured it was easiest to just jump off the note I wrote to myself in Google Keep after the idea popped into my head:
That might not make so much sense (yet), but like any good note it captured enough of the concept that I remembered what I was thinking when I wrote it. I jotted it down as I was prepping for a webinar I did last week offering up some predictions for marketing in 2019. I was getting worked up (as I’m wont to do) about how much it bugs me when everyone in marketing talks about AI as if they have any idea what it really means or the implications.1 Someone asked why it bothered me so much and my answer, which kind of just poured out, was that once everyone starts agreeing about something (and saying it endlessly) it becomes less and less meaningful. This is not just some soft definition of the word meaning, though, it literally has less information.
A few months ago I wrote about Claude Shannon and information theory. Shannon wrote a seminal paper in 1948 called “A Mathematical Theory of Communication“. In it he defined the measure of information as, effectively, its unexpectedness (he called it entropy). The more random, the more information. This is precisely what bits measure (you can think of it as the number of yes/no questions it would take to get to the answer). What happens when you compress a photo? You take away the randomness. That’s why otherwise complex surfaces like sky or skin might come to look a bit pixelated: The compression algorithm is constraining the number of hues available in order to bring down the entropy (and therefore the file size) of the whole photo.
What does that mean for marketing buzzwords?
Well, as everyone starts to say the same thing and continue to offer little behind it, it becomes more and more expected and, therefore, starts to carry less and less information. When people layer on top of those buzzwords with real examples or alternative ideas, they return some randomness (and therefore information) to the concept. At their best, marketing contrarians are attempting to breathe some life into words and ideas that have otherwise lost their information content.
I don’t really like to think of myself as a contrarian because I think that often carries with it some notion of being different for the sake of being different (and trolling). Rather, I think if everyone is following one strategy or idea, the value of being the next person to jump on board is incrementally less (especially when that idea is poorly defined/understood). In a way it’s like an anti-network effect.
Back to Hinkie’s letter. It was leaked and provided an amazing view into the psyche of someone who was willing to be a pariah. In it he paints an interesting picture of the connection between contrarianism and traditionalism.
Here he is on contrarianism:
To develop truly contrarian views will require a never-ending thirst for better, more diverse inputs. What player do you think is most undervalued? Get him for your team. What basketball axiom is most likely to be untrue? Take it on and do the opposite. What is the biggest, least valuable time sink for the organization? Stop doing it. Otherwise, it’s a big game of pitty pat, and you’re stuck just hoping for good things to happen, rather than developing a strategy for how to make them happen.
And on traditionalism:
While contrarian views are absolutely necessary to truly deliver, conventional wisdom is still wise. It is generally accepted as the conventional view because it is considered the best we have. Get back on defense. Share the ball. Box out. Run the lanes. Contest a shot. These things are real and have been measured, precisely or not, by thousands of men over decades of trial and error. Hank Iba. Dean Smith. Red Auerbach. Gregg Popovich. The single best place to start is often wherever they left off.
Let’s bring it back to buzzwords.
So basically Hinkie’s argument is that the most appropriate way to be a contrarian is to also be a traditionalist: To be a respectful student of the underlying principles while also constantly probing and questioning whether they still make sense. One of the things that surprises me about the marketing industry is how often people miss this tradeoff. In an attempt to play the contrarian they shun traditional wisdom, but at the same time they repeat empty phrases and approaches at every conference that will let them on stage.
I actually think one of the reasons Byron Sharp’s book How Brands Grow has picked up as much steam as it has is because it strikes a good balance between these things. It’s a contrarian take (loyalty shouldn’t be a goal because it’s an outcome) but at the same time it’s deeply rooted in some traditional marketing ideas (marketshare, reach, and creativity to name three). This is a tough balance to strike, but when someone hits the spot is has the opportunity to really resonate.
Unfortunately, most of the time the industry misses the market by a lot. What we end up with a bunch of anti-historical/anti-intellectual slogans that get repeated ad-infinitum. It’s lots of words and little information.
Here’s the notes I had for the question: “Let me start by saying that I predict in 2019 marketers will continue to talk about AI and ML interchangeably with no idea what the words mean. (I’m particularly salty about this.) I would broadly see we will continue to see ML become more available as different kinds of wrappers are made available that enables folks to use it in more of their everyday work. This seems to be some of what Microsoft and Google are doing with smart integrations into their work suites. In general, my take on AI/ML is it’s a classic case of Amara’s law, “We tend to overestimate the effect of a technology in the short run and underestimate the effect in the long run.” In the short term, these things aren’t going to be writing copy and, anyway, that’s not that big a deal. In the long term, the promise of ML is data modeling and coding written by computers, not people. That’s definitely not a 2019 prediction, but it’s the road we’re going down.”↑
I recognize that the word/idea transformation belongs in the buzzword bucket, but if you read about Hinkie and what he did I think it’s a fair use of the word with real meaning. He was a heretic who questioned the most fundamental law of professional sports (“you play every game to win”) and rewrote the path to building a championship contender.↑
It’s been awhile since I did a Remainders posts so I figured I’d throw one together. In theory it’s all the other stuff I didn’t get a chance to blog about. In reality, it’s pretty much everything I’ve been reading that isn’t about mental models/frameworks (and even some of that). You can find previous versions filed under Remainders and, as always, if you enjoy the writing, please subscribe by email and pass around.
Let’s start with some books. Here’s what I’ve read in the last three months (in order of when they were read):
Countdown to Zero Day(Kim Zetter): As far as I know this is the definitive book on Stuxnet, the digital weapon that targeted the Iranian nuclear facility at Natanz.
Complexity: A Guided Tour (Melanie Mitchell): Easily one of my favorite books of the year. I’ve read lots about complexity theory, but nothing that pulled all the various strings together so well. (This also helped send me down a deep physics rabbit hole that I’ve yet to emerge from.)
A Brief History of Time (Stephen Hawking): If you find yourself in a physics rabbit hole, this seems like something worth reading …
Dreamtigers (Jorge Luis Borges): I read about this in the Borges interview book. He basically explained that his publisher asked for a book and so he collected a bunch of poems and stories that were sitting around his house and hadn’t been published and stuck it together.
Okay, onto some other reading, etc. …
This Wired piece about the possibility of a coming “AI cold war” has two particularly interesting strings in it: One is a fundamental question about the nature of technology and its relationship with democracy (put simply: is the internet better structured to support or defeat democratic ideals) and the other is about how China (and the US) will use 5G as a power play (“If you are a poor country that lacks the capacity to build your own data network, you’re going to feel loyalty to whoever helps lay the pipes at low cost. It will all seem uncomfortably close to the arms and security pacts that defined the Cold War.”)
Benoît Mandelbrot (of fractal fame) is apparently responsible (at least in part) for the introduction of passwords at IBM. From When Einstein Walked with Gödel (which I’m reading now), “When his son’s high school teacher sought help for a computer class, Mandelbrot obliged, only to find that soon students all over Westchester County were tapping into IBM’s computers by using his name. ‘At that point, the computing center staff had to assign passwords,’ he says. ‘So I can boast-if that’s the right term-of having been at the origin of the police intrusion that this change represented.'”
Also from the same book, the low numerals are meant to be representative of the number of things they are. Since that makes no sense, here’s the quote from the book: “Even Arabic numerals follow this logic: 1 is a single vertical bar; 2 and 3 began as two and three horizontal bars tied together for ease of writing.”
A Rochester garbage plate “is your choice of cheeseburger, hamburger, Italian sausages, steak, chicken, white or red hots*, served on top of any combination of home fries, french fries, baked beans, and/or macaroni salad.”
Rahimi believes contemporary machine learning models’ successes — which are mostly based on empirical methods — are plagued with the same issues as alchemy. The inner mechanisms of machine learning models are so complex and opaque that researchers often don’t understand why a machine learning model can output a particular response from a set of data inputs, aka the black box problem. Rahimi believes the lack of theoretical understanding or technical interpretability of machine learning models is cause for concern, especially if AI takes responsibility for critical decision-making.
Uber’s business plan, like that of so many other digital unicorns, is based on extracting all the value from the markets it enters. This ultimately means squeezing employees, customers, and suppliers alike in the name of continued growth. When people eventually become too poor to continue working as drivers or paying for rides, UBI supplies the required cash infusion for the business to keep operating.
West calls his struggle the right to be a “free thinker,” and he is, indeed, championing a kind of freedom—a white freedom, freedom without consequence, freedom without criticism, freedom to be proud and ignorant; freedom to profit off a people in one moment and abandon them in the next; a Stand Your Ground freedom, freedom without responsibility, without hard memory; a Monticello without slavery, a Confederate freedom, the freedom of John C. Calhoun, not the freedom of Harriet Tubman, which calls you to risk your own; not the freedom of Nat Turner, which calls you to give even more, but a conqueror’s freedom, freedom of the strong built on antipathy or indifference to the weak, the freedom of rape buttons, pussy grabbers, and fuck you anyway, bitch; freedom of oil and invisible wars, the freedom of suburbs drawn with red lines, the white freedom of Calabasas.
This hits close to home: Your coffee addiction, by decade. “‘No sugar,’ you declare. ‘I take it black.’ Shoot a side-eyed glance at that kid over there with his blended-ice drink—amateur hour. Sorry they don’t serve Shirley Temples, geez.”
On the podcast front, I’ve been enjoying Real Famous, which features interviews with ad people (many of whom are my friends). Paul Feldwick, author of the awesome book Anatomy of a Humbug, is an excellent listen.
Multitasking, in short, is not only not thinking, it impairs your ability to think. Thinking means concentrating on one thing long enough to develop an idea about it. Not learning other people’s ideas, or memorizing a body of information, however much those may sometimes be useful. Developing your own ideas. In short, thinking for yourself. You simply cannot do that in bursts of 20 seconds at a time, constantly interrupted by Facebook messages or Twitter tweets, or fiddling with your iPod, or watching something on YouTube.
You could say the trouble for Rodger started when, around puberty, he began to know—and, in writing, recite—the first and last names of every boy he considered a sexual competitor, while at the same time referring to girls almost always collectively. Girls. Pretty girls. Pretty blond girls. Only three girls (or perhaps, by this time, women) are listed by name in My Twisted World, vis-a-vis dozens of boys (I’m not including family members). By the end of his writing and life, he’s failed to distinguish between any groups of humans at all, to the point where he considers his 6-year-old brother yet another budding Romeo who, because “he will grow up enjoying the life [Rodger has] craved for,” must die. “Girls will love him,” Rodger says. “He will become one of my enemies.” Rodger begs our most individuating question—“why don’t you love me?”—by proving himself repeatedly unable to individuate another. In erotic coupling, the ego finds relief in its equal. But had Elliot Rodger ever found his equal and opposite in another human being, he would, by all indications, have been repulsed. Reading him, I kept remembering Rooney Mara’s kiss-off in The Social Network: “You are going to go through life thinking that girls don’t like you because you’re a nerd.1 [Or short. Or half-Asian. Or bad at football, or not a real ladies’ man, or somehow else disappointing to the ur-dads of America.] And I want you to know, from the bottom of my heart, that isn’t true. It’ll be because you’re an asshole.”
Chunking was originally conceptualized in the groundbreaking work of Herbert Simon in his analysis of chess—chunks were envisioned as the varying neural counterparts of different chess patterns. Gradually, neuroscientists came to realize that experts such as chess grand masters are experts because they have stored thousands of chunks of knowledge about their area of expertise in their long-term memory. Chess masters, for example, can recall tens of thousands of different chess patterns. Whatever the discipline, experts can call up to consciousness one or several of these well-knit-together, chunked neural subroutines to analyze and react to a new learning situation. This level of true understanding, and ability to use that understanding in new situations, comes only with the kind of rigor and familiarity that repetition, memorization, and practice can foster.
The computer takes a reading from a Geiger counter that measures radiation in the surrounding air, specifically the radioactive isotope Americium-241. The reading is expressed as a long number of code; that number gives the generator its true randomness. The random number is called the seed, and the seed is plugged into the algorithm, a pseudorandom number generator called the Mersenne Twister. At the end, the computer spits out the winning lottery numbers.
If you haven’t heard the Google Duplex calls, go have a listen. Some interesting comments from Twitter:
Jessi Hempel: “Reading about Google’s Duplex: Design is a series of choices, and creating voice tech designed to let humans trick other humans is a choice humans are making, not an inevitable consequence of technology’s evolution.”
Stewart Brand: “This sounds right. The synthetic voice of synthetic intelligence should sound synthetic. Successful spoofing of any kind destroys trust. When trust is gone, what remains becomes vicious fast.”
The New York Times’s Weinstein report was a believability project years in the making: it systematized abuse, turned it into a pattern your eye could follow. There were interviews, emails, audio recordings, legal documents; facts were double- and triple-checked. But its paradoxical consequence was to set the bar far too high for every subsequent story whose breaking it had made possible. What’s a little masturbation between friends when the king of Hollywood kingmakers had employed former agents of the Israel Defense Forces to silence his accusers? In one final act of gaslighting, Weinstein made all other abuse look not so bad and all other evidence look not so good.
Annnnnd here’s my 10th blog post of the month. Hit my goal. (Might even make it to 11 if I have a burst of inspiration.) Thanks again for reading and encouragement. I’m going for 10 again in May. As usual, feedback welcome and you can subscribe by email here (for those of you reading this via email, thanks and sorry about the wasted words, it just emails exactly what I put on the web).
Black infants in America are now more than twice as likely to die as white infants — 11.3 per 1,000 black babies, compared with 4.9 per 1,000 white babies, according to the most recent government data — a racial disparity that is actually wider than in 1850, 15 years before the end of slavery, when most black women were considered chattel. In one year, that racial gap adds up to more than 4,000 lost black babies. Education and income offer little protection. In fact, a black woman with an advanced degree is more likely to lose her baby than a white woman with less than an eighth-grade education.
By rendering a not-too-distant future, Kubrick set himself up for a test: thirty-three years later, his audiences would still be around to grade his predictions. Part of his genius was that he understood how to rig the results. Many elements from his set designs were contributions from major brands—Whirlpool, Macy’s, DuPont, Parker Pens, Nikon—which quickly cashed in on their big-screen exposure. If 2001 the year looked like “2001” the movie, it was partly because the film’s imaginary design trends were made real.
The show offers a clever finger trap for critics. Call a hit dangerous and you imply that it’s really quite sexy. And, in fact, the seventh episode, which I won’t spoil, pulls a daring switcheroo, one that may offer a new lens through which to interpret Roseanne’s behavior. It’s not enough. The reboot nods at complexity without delivering—there are good people on many sides, on many sides. If you squint, you might see the show’s true hero as Darlene (Sara Gilbert), a broke single mom forced to move in with that charismatic bully Roseanne. But, if that were so, we might understand Darlene’s politics, too. We’d more fully feel her pain and also that of her two kids, transplanted to a place they find foreign and unwelcoming.
This is where the promise of artificial intelligence breaks down. At its heart is an assumption that historical patterns can reliably predict future norms. But the past—even the very recent past—is full of words and ideas that many of us now find repugnant. No system is deft enough to respond to the rapidly changing varieties of cultural expression in a single language, let alone a hundred. Slang is fleeting yet powerful; irony is hard enough for some people to read. If we rely on A.I. to write our rules of conduct, we risk favoring those rules over our own creativity. What’s more, we hand the policing of our discourse over to the people who set the system in motion in the first place, with all their biases and blind spots embedded in the code. Questions about what sorts of expressions are harmful to ourselves or others are difficult. We should not pretend that they will get easier.
On the other end of the sporting spectrum, the Times got a hold of tapes from a meeting between players and owners and I can’t imagine it making the NFL look worse. Here’s a small example from Buffalo Bills owner Terry Pegula: “For years we’ve watched the National Rifle Association use Charlton Heston as a figurehead … We need a spokesman.” These guys are such bad news.