All Posts

Heuristics and Empirics, Natural Frenemies

A simple question -- How do you know when you’re right? – led me down a long and treacherous path of self-discovery. The truth is that it is very easy to think that content models are correct, vocabularies complete, and process flows comprehensive and strategically appropriate, but very hard to actually know. (And frankly, this is not good for business!)

I began exploring the patterns of both development and validation – how models are created and then tested – and found a deceptively simple duality: heuristics vs. empirics. Heuristics can be thought of as the mental or creative model, based on what feels like gut instinct but is actually the wealth of tacit knowledge accumulated over time. Heuristic thinking is disguised as “genius” and “intuition,” but it’s also the only way to predict and define a future state that doesn’t yet exist outside the imagination. Empirics, in contrast, are all about the current reality, the often complicated truth. Empiricists are natural guides through the intricacies of hard data, analytically synthesizing the details and behavior of actual data and actual content consumers (users). Wherever the heuristic strategist dreams of an elegant solution, the empirical realist smashes it with measurable fact.

And so heuristic and empirics are both natural partners and natural enemies. You might even say frenemies. And their stadium is one of current state. Heuristic approaches are about advancing current state to something faster, more elegant and more efficient, whereas empirical approaches are about better understanding the current state, discovering patterns of strength and weakness not through brainstorming but simple observation.

Let’s take the example of a controlled vocabulary, such as a list of product categories. If a florist wanted to build a list of words that broadly describe collections of product types, it could approach the problem both heuristically and empirically. Heuristically, a few experienced people can gather in a room and collaborate on how they think about florists products. There are things that grow, like seeds and plants, and things that don’t, like soil and fertilizer and tools. Plants themselves can then be divided into types of plants, such as flowering and nonflowering, edible and inedible, aromatic and nonaromatic, native and nonnative. This top-down brainstorming is indicative of heuristic work.

Alternatively (or simultaneously), a hands-on inspection of all of the florist’s current products, as well as any products they expect to sell in the near future and those sold by competitors, can uncover some interesting perspectives. There are products bought by amateurs or professionals, in bulk or in individual units, during the summer months or all year round, from brand-name vendors or with indifference to source, paired with other products or in isolation. The key to this bottom-up approach is access to data: catalogs (which may already have structure that’s helpful or misleading), purchase records, seasonal data, search queries, behavioral observations. Rarely are empirical data “pretty,” but a smart visualization can bring this knowledge quickly to the light.

Of course, all of this doesn’t yet answer the original question, How do you know when you’re right? For this answer, you’re going to have to wait for my next post. In the meantime, recognize that this question must be answered with a question: Do you mean heuristically or empirically? Because clearly the answer is something like “a little bit of both.”

For more on this topic see, Using CRANIUM for Empirical Testing.

Recent Posts

Knowledge Engineering, Knowledge Management and AI Assistants

AI assistants are a form of cognitive AI. They go by many different names. The different terms used to identify AI assistants are described below. The role of AI assistants is to support customers or employees as they carry out tasks. In some cases, for repeatable actions with unambiguous outcomes, a chatbot or AI assistant can fully automate interactions. In other cases, when the answer requires judgment and human expertise, the AI assistant can surface the necessary information to help the human solve a problem – whether for a customer self-service app or a customer service agent supporting an end user task. These virtual job assistants are designed to help customer support staff be more efficient, accomplishing the task more quickly and accurately with a consistent outcome.

[Earley AI Podcast] Episode 22, Peter Voss

In this episode, Seth and Chris talk with Peter Voss,  Founder, CEO, and Chief Scientist at AGI Innovations & Aigo.ai.

[Earley AI Podcast] Episode 21, Dan Turchin

Guest: Dan Turchin, CEO & Founder, PeopleReign