In 2023 we introduced Future Dispersion Protocol, or FDP, as a way to study the changes in society, technology, business leadership, and government that would be needed a century from now if more people were to live better lives. The work turned on two questions: what better means in a particular locality, physical or virtual, and what individual and small-group changes would have to occur to achieve it.
From Stance to Practice
Those questions still hold. What follows is how a leadership team can use them. A protocol has to produce definitions other people can contest, paths that can fail a check, and near-term work with an owner. This follow-up outlines how we use FDP as an operating practice for the organizations we advise.
Who the Work is For
FDP is not a blueprint for society. Detailed social planning has a poor record. The useful unit is a locality a leadership team can actually change: a company, a plant network, a city function, a hospital system, a standards body, a platform.
The horizon is about 100 years because that is long enough for infrastructure, skills, and institutions to turn over, and short enough that decisions made this decade still bind the people who inherit them. The people who matter are the living and their descendants. We receive capital, knowledge, debts, and locked-in systems from those who came before. The obligation is to use that inheritance without casually destroying the options the next three generations will need.
What Better has to Mean
Better cannot remain an atmosphere. Before any future-mapping starts, a locality writes a better-statement that names five things.
Whose welfare counts. Employees, customers, residents, suppliers, future hires, and people outside the locality who absorb the side effects. If an affected group is unnamed, the statement is unfinished.
Which dimensions move. At minimum: health and safety, material security, useful competence, agency over work and data, time, ecological load, and the ability to notice and reverse a bad path.
Trade-offs. A gain that requires a silent loss elsewhere is a transfer, not an improvement. Say what you are willing to give up.
Distribution. Averages hide concentrated harm. “Better for most” only holds if the worst-off group in the locality is not paying for the headline result, or if that cost is explicit and offset.
Reversibility. A future that cannot be exited without collapse does not qualify as better, even if it is comfortable for the current occupants.
A statement that cannot survive those tests is a slogan. Stop and rewrite it.
The Six Artifacts
FDP has six written artifacts. If they do not exist, the team is discussing the future. It is not running the protocol.
-
- Horizon frame. A date about a century out, a named locality, and the constraints that will still bind: energy, skills, regulation, installed equipment, customer lock-in.
- Better-statement. The five-part definition above, written so a skeptical operator can disagree with it in public.
- End-state sketches. Three to five internally consistent pictures of the locality at the horizon, including the version where better is only partly met and the version where it is met in a way that creates a new failure. Branching is the work, not an extra.
- Backcast. From each sketch, work backward in 15-25 year bands until you reach the present. Each band must name the capability, standard, and skill that have to already exist, and the lock-in that must not exist.
- Forecast check. Separately project forward from today using only trends you can defend. Where the forward path and the backcast diverge, tag the gap as already underway, contingent on a decision this decade, blocked by an identifiable interest, or likely to go off the rails.
- Near-term portfolio. Convert surviving requirements into 12-36 month actions with owners, costs, and kill criteria. Review when a leading indicator breaks, not when the planning calendar says so.
Those six objects are the FDP model. Software is optional. The model is the paper trail and the rule for revising it.
What this Means for Business
Technology and the search for better cause each other. The useful distinction is not which came first. It is which mechanism you are in:
-
- Demand pull: a group trying to live or work better funds a tool.
- Supply push: a tool appears because it is possible, then rewrites what better means.
- Loss and lock-in: a capability exists, then cannot be taught, inspected, or replaced.
The Antikythera mechanism is evidence for the third. It was a geared analog computer from the Hellenistic world, accurate enough that comparable mechanical complexity did not reappear in Europe for many centuries. We know it because a ship sank. The design tradition around it did not survive. The lesson for a company is narrower than alternate history. If the method that makes you valuable cannot be taught after the current team leaves, or cannot be run without a single vendor, you are already practicing a small version of that loss.
The commercial questions FDP forces this year are practical:
-
- If this product became infrastructure, who could not exit it?
- Which skills does it atrophy, and are those skills load-bearing in the better-statement?
- What would have to be standardized by 2040 for the 2126 sketches to remain reachable?
- Which future market does the plan assume that the forecast check does not support?
Where Applied Artificial Intelligence does the Work
Artificial intelligence is not the protocol. It is how a midsize organization can run the protocol at a useful speed.
Used well, it stress-tests a better-statement against groups the draft ignored, generates end-state sketches that obey stated constraints and then tries to break them, checks whether a backcast assumes a capability the previous band never produced, monitors a short list of leading indicators, and translates a surviving requirement into implementation options ranked by switching cost and reversibility.
Used badly, it produces fluent futures that cannot be audited. Models do not get to choose the moral content of better, bury a trade-off in a summary, or treat a generated scenario as evidence.
That is how FDP connects to the rest of our work. Use-case selection, data readiness, operating models, and value tracking still matter. FDP adds a test those artifacts often lack: will this initiative still be a good bet if the locality’s definition of better, and the lock-ins it is creating, are taken seriously?
What We Recommend
A method that never produces recommendations is unfinished. These are not predictions. They are present requirements that survive a first pass of backcast and forecast check in many of the industrial and commercial localities we see.
This decade. Build the competence to specify, audit, and refuse a system before scaling the system. Prefer contracts, data schemas, and workflows that can be unwound. Track floor conditions (time to service, preventable harm, vendor concentration, time to replace a critical skill) not only averages.
The next 25 years. Design for error-correction: portable records, independent evaluation, and procurement that rewards switching. Train for the tasks that remain after automation, not for job titles that will not.
The century sketches. Descendants inherit systems. The highest-leverage act is often not a new vision. It is refusing a lock-in that collapses option value. Agencies and firms both need published better-statements for irreversible commitments, and a way to kill a path when the forecast check fails.
If a recommendation cannot be tied to a named locality, a better-statement, and a divergence tag, it is not an FDP recommendation.
The operating answer is smaller and harder: write the six artifacts, put an owner on the first reversible action, and treat every AI investment as a choice about which futures you are still allowed to reach.
For more information about how our work in Applied Artificial Intelligence can help you, please contact us HERE.