LT Wilson · provenance

Knowledge Harvesting®: 37 Years to Scale

Where the method came from, and why it scales now

Knowledge Harvesting did not begin as a product. It began with a problem I couldn't stop seeing: people with deep, valuable knowledge are rarely invited to share it — and when they are, the knowing has become automatic, and automatic knowledge is hard to say.

1989, EOR: Establish, Organize, Represent.

The work begins in a master's thesis called EOR: Establish, Organize, Represent — The Foundation for Occupational Applications. Its dialogue subsystem, Tutor, was a structured-questioning engine: the direct ancestor of everything that followed.

1992, Commercialized EPSSs, aka, LearnerFirst.

The thesis became commercial software. In 1992, one year after Gloria Gery defined electronic performance support systems in her book, LearnerFirst had them on the market, sold under the promise "Learn it, do it, document it, all at the same time." LearnerFirst delivered the applications to over 4,000 organizations: Benchmarking, Process Management, Implementing ISO 9000, Work Profiling, and more. This provided an early operational demonstration that structured questions could carry expert knowledge to people who needed it.

1996, Knowledge Harvesting in the Field

From the mid-90s on, the method ran as Knowledge Harvesting®: hundreds of expert projects with top performers across industries, each with four to twenty elicitation sessions, thousands of sessions in all: troubleshooting nuclear reactors, designing chemical plants, coordinating pandemic operations, writing patent claims, steering drill strings, capturing what a nurse's diagnostic instinct actually does. Clients ranged from Intel, Chevron, and Dow to Abbott, GE, and the Florida Department of Education. The paper trail spans three decades of publications and conference work, from 1997 to 2019.

Why it didn't scale before

Human-to-human elicitation worked. But it was labor-intensive: one practitioner (often, a team), one expert, discussions. The method's ceiling was my calendar. What was missing was an instrumented way to run the questioning, something that could interview, structure, inspect, and organize expertise without the human practitioner.

8 and 96

That is what the instruments are. 8 showed which knowledge types were showing up in a conversation and, after it absorbed Protocol, ran the work loop. I retired it in September 2026 because it had grown too many features; its successor, on the home page, keeps four. 96 is the same engine at professional resolution, in active build. The capture surface became repeatable without requiring me to lead every session. That practical change makes broader use possible.

The method deploys at knowledgeharvesting.com. The instruments live here.
© 2026 Larry Todd Wilson. All Rights Reserved. · ltwilson.com