Effective leaders consistently read. To keep up the habit, every month we pull a book from the Recharted Territory reading list for inspiration. The book for April was “Automate This: How algorithms took over our markets, our jobs, and the world” by Christopher Steiner. It’s a collection of case studies about how algorithms have impacted different industries.
Steiner’s goal with the book was to raise public awareness of the role that algorithms now play in our lives. He doesn’t comment much on what he thinks it means for society, but cautions that industries will rapidly change. He also predicts that people who know how to create algorithms will continue to be valuable.
At the most basic level, algorithms take inputs from the world, follow a set of instructions, and produce an output. More advanced bots can work with thousands of inputs and functions to achieve results faster and often with more accuracy than humans. They are dynamic, and can learn and improve by themselves.
Algorithms will continue to disrupt industries, and organizations without plans to apply them will fall behind. The gap between the algorithmic haves and the have-nots could be similar to other technical revolutions, like not having access to clean water, electricity, or the internet. Some solutions are available for purchase, others require in-house expertise, and some gains could be achieved with a simple mindset shift from status quo.
As portfolio managers, looking at all of the activities and resources we’re investing in within our organization, underestimating the power of algorithms is short-sighted. But how exactly do we elevate the role of algorithms in our portfolio? Steiner’s review of algorithms through history inspires some insights and actions that we can apply today.
You already use algorithms
Computers might have taken over most of the execution, but algorithms are inherently human. Algorithms are the structure we apply to decision making and predicting the consequences of taking an action. We identify patterns in our lives and work, building on what we’ve seen before to solve problems and create new patterns.
Humans applied algorithms long before computers existed (the first recorded algorithm is from 2500 BC). Computers help create more sophisticated algorithms, but the first step was to look for structure within a concept.
As humans, we calculate some algorithms without realizing it. Any algorithm starts simple. If this input is found, then do that. Gottfried Leibniz realized that by stringing together large amounts of simple decisions, very complicated decisions could be made. Aided by math, science, and economics, we could build increasingly sophisticated and accurate algorithms.
If you’re not applying an algorithmic perspective to your own business to improve, automate, or standardize your processes then you’ll fall behind fast. Recognizing where algorithms are used in your business and what they are is a first step toward identifying which are candidates for higher level automation. Standard operating procedures and call scripts are common sources. This can be a challenge for creatives or consultants who don’t think that they follow patterns or are reluctant to codify them. But some professionals end up seeing algorithms as an asset. Another tool to speed up decision making or reveal new insights that complement their skill set.
Goals drive the creation of algorithms
Whether to make a profit or compose an opera, the people that Steiner profiled created algorithms to accomplish a goal. To create something new, solve a problem, predict a future outcome, or make a better decision than the competition. The statement of a goal helps to focus the work required to build, test, and deploy a new algorithm.
Embed algorithms in business models
An algorithm by itself is a tool, not a business model. Algorithms that helped their creators become a leader in their industry were embedded in business models. They often contributed to a value proposition of higher accuracy, speed, or quality. People used the information they gathered via algorithms to make decisions on purchases, build new content, or reduce risks. They designed and sold tools that used algorithms to solve customer problems. Incorporating algorithms into your portfolio means treating them as a component of a business model, not a standalone technical capability.
The model matters so leverage science, math, and other academic journals for inspiration
The value of many algorithms is predicting what will happen or closely reproducing something in the real world. Depending on the problem, the choice of model can matter greatly.
Steiner shares a story about the CIA and predicting the outcomes of geopolitical situations. Algorithms, built by Bruce Bueno de Mesquita and based on game theory, ended up being right twice as often as the CIA analysts. The difference was that Bueno de Mesquita’s algorithms were considering what the major players would likely do given their interest in and influence over a decision. In contrast, a standard analyst would absorb as much information as they could about the background and context of the issue and make an informed prediction. The game theory model required fewer data inputs, and the relative accuracy of the results speaks for itself.
When David Cope was working on an algorithm to help compose an opera, he noticed that the resulting music lacked the “vitality” of pieces composed by humans. By instructing the bot to deviate from its normal decisions every once in a while, it created music that was more surprising and interesting. The wrong model can impact results in any field. Some dating site algorithms don’t deliver a long-term match because they exclude factors that cause relationships to fail, like how people communicate and solve problems together.
Companies looking to leapfrog the competition didn’t just follow others in their industry, they read scientific studies and laws published in academic books and journals. Scientists are dedicated to uncovering patterns and laws in the world around us, which can be leveraged in tools to run more effective organizations. A strong portfolio strategy for algorithms would include time for reviews of scientific literature and exploring ways to apply them. At the very least, monitor trends and products that apply algorithms to complete common tasks.
Prioritize access to large amounts of quality data
The core instructions are a challenge to design, but the inputs to the algorithm are also important. Emmy, Cope’s composer algorithm learned by looking at pieces composed by Bach. Financial bots read information from a variety of sources, including specially formatted Bloomberg news reports. The amount and quality of the data feeding into the algorithm helps the bot figure out what patterns are worth emulating. Reviewing the data you have access to, defining the result you’re striving for (ex. replicate something from the past or create something unique), and filling in the data gaps should all be part of your roadmap.
Don’t forget compute power
The software is just one piece of applying algorithms. In some industries (such as finance) where algorithms are similar, the fastest bot wins. For many businesses speed might not make a difference but don’t forget to factor in cloud infrastructure costs in your plan.
Evolve human systems as you go
What happens when bots create music and movie scripts, win sales, diagnosis illnesses, argue cases, and set prices all by themselves? Who is the “author” of the intellectual property? Who is held accountable for the outcomes? The creator of the algorithm, the algorithm, or the person who runs it? What does that mean for human job descriptions, recognition, commissions, and law enforcement?
Designing your business for algorithms means more than applying the algorithm itself. Like any other technical upgrade, a full release means adding management, training, and recruiting projects to your roadmap as well. And in some cases, it means contemplating complicated legal questions.
Prepare for the bot takeover today
Steiner outlined the impact of algorithms on society, especially since the digital revolution. It may take a few more years for algorithms to sweep through all industries but the proactive companies are preparing for that future today. It’s time to ask the “what if” questions. What could your business look like if algorithms did everything? How will you deliver value? How will the needs of your customers change? What will your employees be doing? Which skill sets would you need and how can you train, recruit, or contract to acquire those skills? If planning, training, recruiting, and upgrading your systems aren’t in your portfolio roadmap, then you might be struggling in the near future.
Does your organization explicitly account for algorithms in your portfolio? How do you incorporate them?