The primary goal outlined in the memorandum is to “restore fairness” in these trade relations through tariff equalization—meaning the United States would impose reciprocal tariffs on imports from countries with higher rates than those in the United States. Additionally, the memorandum addresses other nonreciprocal practices, including “unfair, discriminatory, or extraterritorial taxes” like value-added taxes; nontariff barriers, subsidies, and “onerous regulatory requirements on U.S. businesses abroad”; currency devaluation, wage suppression, and other “mercantilist policies” that disadvantage U.S. companies; and “any other practice that . . . imposes unfair restrictions on market access or creates structural obstacles to fair competition with” the United States—providing the administration much leeway in assessing what constitutes unfair trade practices.

Be the first to comment »

Traditionally in the West we define knowledge as a justified true belief. But the experience of knowledge usually also requires understanding and a framework of connected pieces of knowledge.

Guess what machine learning lacks: understandability and a framework from which its statements of knowledge spring.

We might want to say that therefore ML doesn’t produce knowledge. But I think it’s going to go the other way as AI becomes more and more integral to our lives. AI is likely to change our idea of what it means to know something…

Continued at the Peter Drucker Forum

Be the first to comment »

Verizon three days ago sold Yahoo and AOL for a measly $5B. 

The “measly” is not sarcastic. Twenty years ago, Yahoo was worth $125B. Verizon bought Yahoo in 2016 for $4.8B. AOL was once worth $200B, but Verizon bought it in 2015 for $4.4B. Which means Verizon lost $4.2B in total in the sale of both companies. 

The private equity firm they sold it to, Apollo, will do whatever it has to in order to make back their money:

Under Apollo, Verizon’s former media properties will be challenged to grow and become profitable in order to attract yet another sale or exit down the road.

If Yahoo and AOL failed under Verizon, there’s little reason to think they’ll succeed under new management that wants to resell them. As of 2017 there are  2.3 million people still using aol.com as their email address, and that number today includes celebrities such as Tina Fey, Steve Carell and Sarah Silverman. Still, an email user base of 2.3M is unlikely to result in the billions of dollars Apollo would have to make off of it. (I am not wise in the ways of billion dollar businesses, though. If only!)

In short, it’s time to think about moving away from AOL.com. You can, of course, have two email addresses at once, and many email providers will  automatically forward your AOL email to your new address. That means that email sent to your AOL.com address will automatically show up in your new email’s inbox. (Here’s how for Gmail.)

Good luck cutting the emotional cord to a pre-Web Internet provider who most of us thought went away twenty years ago.

1 Comment »

My favorite Einstein quote is the one that says, it is the theory that determines the data.  ;-)

So writes someone to a mailing list I’m on. I couldn’t find a source for that, but it might derive from this Einstein quotation:

Whether you can observe a thing or not depends on the theory which you use. It is the theory which decides what can be observed.

 Unification of Fundamental Forces (1990) by Abdus Salam ISBN 0521371406, page 99, related by Heisenberg.

(By the way, my quick Internet “research” turned up no reliable sources for a related quotation often attributed to Marshall McLuhan: “I wouldn’t have seen it if I hadn’t believed it.” It sounds like him, though.)

But no matter. What I really want to talk about is how the Einstein quotation applies to marketing:

Messages create markets.

This is I think quite literally the case in the pre-personalization view of markets as corresponding to demographic slices. The particular demographics that are assumed to constitute a market are the ones that are believed to be susceptible to the same broadcast message. If white adolescents are thought to respond to “Be a Pepper”, then they are a market. If suburban dads are susceptible to “Come alive with Subaru!”, then they are a market. If short baseball fans are not susceptible to the same message, they are not a market.

Such markets have no reality other than their susceptibility to a message. If and when (= now) messages can be delivered to, and even designed for, based on particular traits, preferences, and behaviors of individuals,  “markets” ?cease to exist. Since personalization can more effectively play on human susceptibility to non-rational appeals, many of us, including me, are of course disturbed about this development. 

But there is another side of this as well.  As Doc Searls once said, markets are conversations. If customers and users now talk with one another about the things they want to buy and have bought, then markets once again have some actual reality. The fact that these conversations occur in and on networks makes these markets fluid but does not make them less real.

Be the first to comment »

On Feb. 19, 2020, Elliot Noss testified before the Canadian Radio and Television Commission about how to knock Canada off the very top of the list of the world’s most expensive mobile connections. It’s very much worth a read [pdf] or view.

Elliot is the president and CEO of Tucows, and the found of Hover and Ting. These companies are profitable, but they are also driven by Elliot’s commitment to supporting an open Internet … where openness includes open affordable. Ting, for example, provides excellent wireless service at prices that should make the Big Boys blush in shame — although they’d first have to look up “shame” in the dictionary — while also providing what may be the best customer service in the world. Not exaggerating. Hover is also a very excellent Web registrar.

Yes, Elliot is a friend of mine. But one of the reasons I’m so attached to him is that he is so thoroughly decent. He is what my tribe calls a Mensch.

In his testimony, he’s trying to get the Canadian government to support Mobile Virtual Network Operators (MNVOs), as opposed to only supporting “facilities-based providers” that, by definition, “serve a subscriber using its own network facilities and spectrum…” [pdf] The facilities-based providers own the wire or cable going to your house, and they compete on the basis of their coverage. An MNVO (such as Ting) rents access to the physical infrastructure and provides services to customers, competing on factors like price, service, and quality … which is what we customers want. (Yes, Canada supporting MNVOs would open up business opportunities for Elliot personally, but that is not his primary driver.)

Here’s a taste of Elliot’s remarks:

Telecommunication services are infrastructure, just like water, electricity and roads. Think of telephone service provided over copper networks. From their onset they were regulated infrastructure with rate of return economics. When we introduced mobile phone service provided
over the public resource of spectrum it was for making phone calls and was considered a luxury. Today it is primarily for using small computers, that we still anachronistically call “phones”, to consume data. And it is for everyone. Lower income Canadians need access to mobile data just
like other Canadians. Not for “occasional use”. Not at lower data rates. In fact lower income Canadians are the most likely Canadians to NOT have a fixed Internet connection at home.

Telecom is infrastructure. Which leads me to my most heretical point. If telecom is infrastructure, and it is, then the desire for facilities-based competition is misplaced. We do not require facilities-based competition with any other infrastructure. In fact it would seem absurd if we were
talking about it in connection with water or electricity.

I am an Elliot Noss fanboy, and proud of it.

Be the first to comment »

Brink has just posted a piece of mine that suggests that the Internet and machine learning have been teaching companies that our assumptions about the predictability of the future — based in turn on assumptions about the law-like and knowable nature of change — don’t hold. But those are the assumptions that have led to the relatively recent belief in the efficacy of strategy.

My article outlines some of the ways organizations are facing the future differently. And, arguably, more realistically.

Be the first to comment »

Digital Extremes wants to break the trend of live-service games meticulously planning years of content ahead of time using road maps…’What happens then is you don’t have a surprise and you don’t have a world that feels alive,’ [community director Rebecca] Ford says. ‘You have a product that feels like a result of an investor’s meeting 12 months ago.'”

— Steven Messner, “This Means War,” PC Gamer, Feb. 2020, p. 34

Video games have been leading indicators for almost forty years. It was back in the early 1980s that games started welcoming modders who altered the visuals, turning Castle Wolfenstein into Castle Smurfenstein, adding maps, levels, cars, weapons, and rules to game after game. Thus the games became more replayable. Thus the games became whatever users wanted to make them. Thus games — the most rule-bound of activities outside of a law court or a tea ceremony — became purposefully unpredictable.

Rebecca Ford is talking about Warframe, but what she says about planning and road maps points the way for what’s happening with business strategies overall. The Internet has not only gotten us used to an environment that is overwhelming and unpredictable, but we’ve developed approaches that let us leverage that unpredictability, from open platforms to minimum viable products to agile development.

The advantage of strategy is that it enables an organization to focus its attention and resources on a single goal. The disadvantages are that strategic planning assumes that the playing field is relatively stable, and that change general happens according to rules that we can know and apply. But that stability is a dream. Now that we have tech that lets us leverage unpredictability, we are coming to once again recognize that strategies work almost literally by squinting our eyes so tight that they’re almost closed.

Maybe games will help us open our eyes so that we do less strategizing and more playing.

Be the first to comment »

Joe McKendrick at Forbes has posted a concise and thoughtful column about
Everyday Chaos, including four rules to guide your expectations about machine learning.

It’s great to see a pre-publication post so on track about what the book says and how it applies to business.

1 Comment »

At the PAPis conference Bas Nieland, CEO and Co-Founder of Datatrics, is talking about how to predict the color of shoes your customer is going to buy. The company tries to “make data science marketeer-proof for marketing teams of all sizes.” IT ties to create 360-degree customer profiles by bringing together info from all the data silos.

They use some machine learning to create these profiles. The profile includes the buying phase, the best time to present choices to a user, and the type of persuasion that will get them to take the desired action. [Yes, this makes me uncomfortable.]

It is structured around a core API that talks to mongoDB and MySQL. They provide “workbenches” that work with the customer’s data systems. They use BigML to operate on this data.

The outcome are models that can be used to make recommendations. They use visualizations so that marketeers can understand it. But the marketeers couldn’t figure out how to use even simplified visualizations. So they created visual decision trees. But still the marketeers couldn’t figure it out. So they turn the data into simple declarative phrases: which audience they should contact, in which channel, what content, and when. E.g.:

“To increase sales, çontact your customers in the buying phase with high engagement through FB with content about jeans on sale on Thursday, around 10 o’clock.”

They predict the increase in sales for each action, and quantify in dollars the size of the opportunity. They also classify responses by customer type and phase.

For a hotel chain, they connected 16,000 variables and 21M data points, that got reduced to 75 variables by BigML which created a predictive model that ended up getting the chain more customer conversions. E.g., if the model says someone is in the orientation phase, the Web site shows photos of recommend hotels. If in the decision phase, the user sees persuasive messages, e.g., “18 people have looked at this room today.” The messages themselves are chosen based on the customer’s profile.

Coming up: Chatbot integration. It’s a “real conversation” [with a bot with a photo of an atttractive white woman who is supposedly doing the chatting]

Take-aways: Start simple. Make ML very easy to understand. Make it actionable.

Q&A

Me: Is there a way built in for a customer to let your model know that it’s gotten her wrong. E.g., stop sending me pregnancy ads because I lost the baby.

Bas: No.

Me: Is that on the roadmap?

Bas: Yes. But not on a schedule. [I’m not proud of myself for this hostile question. I have been turning into an asshole over the past few years.]

Be the first to comment »

Vinny Senguttuvan is Senior Data Scientist at METIS. Before that, he was at Facebook-based gaming company, High 5 Games, which had 10M users. His talk at PAPIs: “Predicting Customers.”

The main challenge: Most of the players play for free. Only 2% ever spend money on the site, buying extra money to play. (It’s not gambling because you never cash out). 2% of those 2% contribute the majority of the revenue.

All proposed changes go through A/B testing. E.g., should we change the “Buy credits” button from blue to red. This is classic hypothesis testing. So you put up both options and see which gets the best results. It’s important to remember that there’s a cost to the change, so the A-B preference needs to be substantial enough. But often the differences are marginal. So you can increase the sample size. This complicates the process. “A long list of changes means not enough time per change.” And you want to be sure that the change affects the paying customers positively, which means taking even longer.

When they don’t have enough samples, they can bring down the confidence level required to make the change. Or they could bias one side of the hypothesis. And you can assume the variables are independent and run simultaneous A-B tests on various variables. High 5 does all three. It’s not perfect but it works.

Second, there is a poularity metric by which they rank or classify their 100 games. They constantly add games — it went from 15 to 100 in two years. This continuously changes the ranking of the games. Plus, some are launched locked. This complicates things. Vinny’s boss came up with a model of an n-dimensional casino, but it was too complex. Instead, they take 2 simple approaches: 1. An average-weighted spin. 2. Bayesian. Both predicted well but had flaws, so they used a type of average of both.

Third: Survival analysis. They wanted to know how many users are still active a given time after they created their account, and when is a user at risk of discontinuing use. First, they grouped users into cohorts (people who joined within a couple of weeks of each other) and plotted survival rates over time. They also observed return rates of users after each additional day of absence. They also implement a Cox survival model. They found that newer users were more likely to decline in their use of the product; early users are more committed. This pattern is widespread. That means they have to continuously acquire new players. They also alert users when they reach the elbow of disuse.

Fourth: Predictive lifetime value. Lifetime value = total revenue from a user over the entire time the the produced. This is significant because of costs: 10-15% of the rev goes into ads to acquire customers. Their 365 day prediction model should be a time series, but they needed results faster, so they flipped it into a regression problem, predicting the 365 day revenue based on the user’s first month data: how they spent, purchase count, days of play, player level achievement, and the date joined. [He talks about regression problems, but I can’t keep up.] At that point it cost $2 to acquire a customer from FB ad, and $6 from mobile apps. But when they tested, the mobile acquisitions were more profitable than those that came from through FB. It turned out that FB was counting as new users any player who hadn’t played in 30 days, and was re-charging them for it. [I hope I got that right.]

Fifth: Recommendation systems. Pandora notes the feature of songs and uses this to recommend similarities. YouTube makes recommendations made based on relations among users. Non-matrix factorization [I’m pretty sure he just made this up] gives you the ability to predict the score for a video that you know nothing about in terms of content. But what if the ratings are not clearly defined? At High 5, there are no explicit ratings. They calculated a rating based on how often a player plays it, how long the session, etc. And what do you do about missing values: use averages. But there are too many zeroes in the system, so they use sparse matrix solvers. Plus, there is a semi-order to the games, so they used some human input. [Useful for library Stackscores
?]

Be the first to comment »