Hinton is quoted as saying, with respect to back propagation, “I don’t think it’s how the brain works”. You can read the full article here. Back Propagation To oversimplify, in Back Propagation, the influence of each neuron is rewarded based on how well it predicts something. Accurate predictions are rewarded with more influence. Bad predictions are punished with less. This is how the machine learns. And there’s a lot of optimism about Back Propagation. It’s really useful and generates fairly predictable machines. As data scientists, we like this. And as data scientists, we should also like what Hinton is hinting at. Kuhn It’s much more likely than not that we’re approaching a local maxima on this thread of research. I’m[…]

An orthodox Software as a Service (SaaS) business is, in part, math that an organization tries its best to manage. Think about all the math that goes into the construction of a typical SaaS firm. At the core there’s some customer with a job: a goal against which the customer wants to make progress. They can have a mathematical representation in a database somewhere. A bunch of technologists write some code, which is all math, and a bunch of creatives take a few photographs, which expresses itself a mathematical representation, and some data is Created Read Updated and Destroyed in a database somewhere, which is all just more math. And it’s all abstracted by yet more math at the processor[…]

The other I likened the process for taking apart a Job To Be Done to taking a part a lobster. There’s a very effective way to decompose any problem with enough energy. And then I watched The Founder on Netflix and admired the McDonald brothers using a classic technique in management science to refine a system on a tennis court. And I loved it. They really refined hamburger and frenched fry delivery. And then this morning I read that Andrew Ng in working on a new coursera course for AI. And I’m thankful for his initiative and optimism. Out of those three threads, this one post. The Assembly Line The assembly line was an American invention for Americans. It could[…]

Earlier in the month, I dined under the space shuttle Endeavour with some of the best minds in marketing science. One mind remarked: “That’s why I bring a glossary with me, oh, you want to do supervised learning? Oh you mean regression? Oh, okay, now we can talk… We’ve been talking to managers about these methods for decades, but it’s just suddenly sexy because it’s all machine learning and deep learning and reinforcement learning.” A lot of the math that underlies much of machine intelligence and artificial intelligence is indeed remarketed marketing science. And, hipsterism aside, the annoyance is understandable. Marketing science started out a bit of a revolt against the Mad Men. Some of the early stories feature post-war[…]

What if Total Addressable Market can’t be estimated accurately? What then? What is Total Addressable Market (TAM)? Total Addressable Market, or TAM, is the number of buyers who are Willing To Pay (WTP) for a solution to a problem they have now, or are Willing To Pay (WTP) your firm instead of the firm they’re currently paying to solve a problem. Why is TAM important? TAM determines the life and death of a firm. The leading cause of startup failure, and perhaps all business failure in general, is the failure to penetrate and/or retain TAM (Including bureaucratic capture and rent-seeking). In this context, I’m concerned about the introduction of a new product into the market in an effort to generate both[…]

The data is imperfect. Judgement is imperfect. Decisions are imperfect. The question isn’t about perfection. It’s about progression. What becomes true if we were to focus on progression? Credit goes to Matt Gershoff for inspiring this post. A remark he made at a recent #miToronto grabbed me. To paraphrase, he said that when you stop obsessing over which model is right or wrong, because all models are wrong by definition, and start focusing on just making it better, you get a lot further. He used the term liberating. And it is. The Data Reality is flawed, and as a direct result, it generates data that is flawed. The machine housed in your skull is a serious piece of technology, but[…]

Bart Gajderowicz delivered a great talk at Machine Intelligence Toronto about how people go through stages in accomplishing a goal [1]. The talk was about homelessness and AI approaches to public policy. I instantly saw a connection to all sorts of tensions that people endure when they set out on a goal. To distill the concept, let’s start off with the idea that people have goals, people have emotions, and that time moves forward. As people make progress towards their goals, their emotions change over time. They start off in a good mood, in a state of uninformed optimism. Then, as negative information overwhelms their ignorance, they enter into a state of informed pessimism. So much negative information builds up[…]

Ben Thompson calls culture the accumulation of decisions. Assume that it’s true. How do decisions at a tech startup come into being in the first place? A startup can be instantiated with the business plan. And if you take a Beinhocker (2006, The Origin of Wealth) approach to it, you may believe that there’s a Library of Smith which contains every single business plan that’s possible. There are trillions upon trillions of potential business plans. And management is pretty much reduced to a machine that is able to execute the plan to generate wealth. Everything that has potential is possible at the beginning and assume competent management. (Image related – a bit esoteric*). In the context of a startup, a[…]

Previously, I argued that you should look at the Q4-2016 VR sales figures closely and then make decisions about whether to jump in. Some figures are in. SuperData Research, a technology research firm, estimated that Oculus had sold 360,000 headsets and HTC 450,000 since their products went on sale in March and June, respectively. Both of those headsets require high-end PCs with powerful processors. The firm estimated that Sony, which began selling a virtual reality headset in October, has sold about 750,000. — NYtimes Jan 8/2017 Those aren’t encouraging install bases. Obligatory Gartner Hype Cycle image: Consolidation is a long ways off. Facebook, has deep pockets and can sustain a long chasm crossing. The legal issues with Zenimax are a distraction. This is[…]

“A study at Ball State University’s Center for Business and Economic Research last year found that trade accounted for just 13 percent of America’s lost factory jobs. The vast majority of the lost jobs — 88 percent — were taken by robots and other homegrown factors that reduce factories’ need for human labor.” – AP Canada’s labour force is around 19.6 million people, of which 18.2 million people are employed. Together, they worked something like 2.4 billion hours that month. In December 2016, something like 1.7 million Canadians worked about 240 million hours in manufacturing.  Roughly. Because of seasonal adjustments and different data at different times. And error. In terms of our working lives in Canada, collectively, manufacturing is about 10% of[…]