Knowledge that compounds.

Mil­it­ary organ­isa­tions renew their staffs every two to three years. With each rota­tion goes a sub­stan­tial share of what the insti­tu­tion knew: its con­text, its rela­tion­ships, and judge­ment acquired at some cost. Suc­cessor teams recon­struct much of it from first prin­ciples, and the cycle repeats.

Suc­cess­ive gen­er­a­tions of soft­ware have not resolved the dif­fi­culty. They have made it faster to pro­duce work; they have not made it easier to retain it.

• A false choice between builders and practitioners

  The present market obliges people to choose. One may become a tech­nical builder, at an increas­ing dis­tance from how the work is actu­ally done, or remain a pro­cess spe­cial­ist without the means to encode that know­ledge. Because the two com­pet­ences seldom reside in the same person, the know­ledge and the system seldom reside in the same place.

• Disposable artefacts

  Headquar­ters still run largely on doc­u­ments, slides and spread­sheets. Even where modern plat­forms are adop­ted, the tools one officer builds are often unin­tel­li­gible to the next, who reverts to a fresh slide deck or begins again on a new system. Effort is expen­ded; little accu­mu­lates.

• No basis for longitudinal assessment

  Where noth­ing accu­mu­lates, there is no con­tinu­ous record of how decisions were reached. Without one, an organ­isa­tion cannot determ­ine whether it is improv­ing or declin­ing. The cog­nit­ive work was never encoded in its sys­tems; it was held in people, and people move on.

• Wisdom and energy without a common medium

  The senior prac­ti­tioner with dec­ades of exper­i­ence can trans­mit judge­ment only as guid­ance, and guid­ance degrades as it is seri­al­ised, sum­mar­ised and relayed, until intent becomes inco­her­ent. The new­comer brings energy and cur­rent skills but little con­text. In the pre­vail­ing envir­on­ment the two work in dif­fer­ent media and talk past one another. Derived Logic offers them a shared medium: exper­i­ence encoded as struc­ture rather than delivered as instruc­tion, and inher­ited dir­ectly by those who follow.

What Derived Logic provides is not another inter­face. It is the onto­logy beneath: a living, formal model of how an organ­isa­tion’s people, pro­cesses, data and decisions relate to one another.

Users con­struct it simply by doing their work: link­ing a task to the require­ment it serves, a decision to its rationale, a product to those who rely upon it. No ded­ic­ated know­ledge engin­eers are required, and there is no sep­ar­ate data pro­ject.

People linktheir everyday work

The system deriveslogic from those links

The next personbegins further ahead

and contributes more, more readily

Every­one draws upon the same under­ly­ing model, yet each person sees only the por­tion that role, clear­ance and need-to-know permit. Releas­ab­il­ity and clas­si­fic­a­tion are prop­er­ties of the struc­ture itself, enforced as the work is done rather than policed after the fact.

An interface can be copied. An ontology built by an organisation’s own people, which compounds with use, cannot.

Many plat­forms offer to encode an organ­isa­tion’s busi­ness pro­cesses in soft­ware. The offer assumes that the pro­cesses are sound. A good number per­sist only because they always have, and to digit­ise them faith­fully is to make their short­com­ings per­man­ent, and faster.

Because know­ledge com­pounds within Derived Logic, the system can do more than pre­serve a pro­cess: it can inter­rog­ate it and pro­pose better ones, employ­ing two modes of reas­on­ing in com­bin­a­tion.

Your laws, your instructions, your caveats

The guard­rails belong to the cus­tomer. Derived Logic fol­lows the laws, dir­ect­ives and national caveats of the organ­isa­tion that employs it; it does not set itself above them as an arbiter, nor sub­sti­tute its judge­ment for the com­mander’s. Noth­ing crosses a legal or clas­si­fic­a­tion bound­ary unac­coun­ted for.

Decisions judged by what was known

Each decision is pre­served together with the data and rules in force when it was taken. It may there­fore be assessed by the stand­ard of its own time, on the basis of what was then known, and sep­ar­ately from how events unfol­ded. This is the dif­fer­ence between learn­ing from the past and judging it with hind­sight.

Each nation oper­ates its own instance, in which its laws, caveats and pro­ced­ures are encoded. Instances can map their onto­lo­gies to one another, so that part­ners under­stand how each other works without any data leav­ing national con­trol.

Nation A

Processes and caveats

Data remains here

mapped structure

Nation B

Processes and caveats

Data remains here

Each nation determ­ines pre­cisely what it shares, down to an indi­vidual table of pro­cess or caveats. A part­ner that holds another’s rule­book knows, during plan­ning and exe­cu­tion alike, how that ally will oper­ate, rather than dis­cov­er­ing a caveat in the middle of an oper­a­tion.

Coali­tion inter­op­er­ab­il­ity is among the most dif­fi­cult and costly prob­lems in modern war­fare, and it most often founders on the demand to share data. This approach makes no such demand. Allies can act as a single force without sur­ren­der­ing either their data or their sov­er­eignty.

Many languages, one structure

Coali­tions work across lan­guages. Con­cepts in the onto­logy are anchored to open, mul­ti­lin­gual ref­er­ence vocab­u­lar­ies, and every cor­res­pond­ence is typed (exact, close, broader or nar­rower), attrib­uted and dated. Part­ners can there­fore estab­lish, at any later point, what a term means in another nation’s usage and why a given map­ping was made.

Our work is groun­ded in prac­ti­tion­ers’ exper­i­ence of cam­paign and oper­a­tional plan­ning in mul­tina­tional headquar­ters, and in a thesis on com­pound­ing know­ledge.

Derived Logic is an Amer­ican com­pany. It is designed from the outset for use by allied and part­ner nations, with whom we are in dia­logue about their require­ments. National sov­er­eignty, national law and national caveats are premises of the design, not later adapt­a­tions.