Path Impact Assessment (PIA)
Consider the consequences of a decision clearly
On hot days, an urban square is almost empty. Old trees give shade, but their roots make some routes difficult. A new canopy could help quickly; a company offers to pay if it can use part of the square for events. Which plan keeps the square accessible and liveable over time? The PIA follows a proposed decision from the present situation to its possible effects.
Preliminary review: who is affected and what changes?
1 · Which paths become stronger, and which become weaker?
Consider each affected path separately. Older residents need reachable shade, trees need soil and water, and the city needs reliable care. Who can shape future use? Which relations become stronger? What can continue and develop? Record gains and losses before choosing a solution.
2 · Which foundations shared by other paths are affected?
Many possibilities depend on a few shared conditions: accessible routes, water, maintenance and public access. Path ethics calls such supporting developments infrastructure paths. If one fails, several others may be harmed. A plan should state who maintains them and what alternatives exist.
3 · Which effects can be corrected?
A bench can be moved later. A mature tree cannot simply be restored after it is cut down. This difference is called reversibility. Trial changes may be suitable for seating. Permanent interventions require stronger reasons and better knowledge beforehand.
Three questions about the intervention
1 · Does the change support or replace the system’s own feedback?
Feedback helps a system notice errors, correct them and adapt. A change can support this ability or replace it with outside control. Taking over one task is not automatically harmful. For AI, name whose ability is being examined: the model, operator, oversight body or the whole arrangement. Fixed rules can help, such as a reliable audit record.
2 · Who loses important conditions, and were they included?
Who depends on water, access, care or another service? Who gains and who loses? Those affected need real influence on the decision. The influence needed also depends on how vulnerable they are. Information alone is not enough. Representation of those unable to speak for themselves needs justification.
3 · Can the intervention be ended without collapse?
Can the intervention change or end without dependent systems or services collapsing? An off switch is not enough. Another viable way to continue may provide an exit. The conditions and effects of the transition need to be examined.
PIA helps people examine a decision; it does not decide alone. Goals and their justification, rights, power and distribution need separate attention. So do evidence and technical, legal and political questions. Sometimes there is not enough knowledge to judge. The procedure still needs practical testing.
AI review: who can still check?
When outward continuity is misleading
If something continues only through outside support while losing its own ability to develop, a persistence illusion may be present. Replaced internal feedback together with no viable exit is a warning signal. The diagnosis still needs to be checked.
PIA as a learning process
A decision can include conditions: who maintains the trees and canopy, how routes remain open and when experience is reviewed. It should name responsible people and a review date. If an assumption proves wrong, there should be an agreed way to revise the plan.
This is a constructed decision case from “The Habitable Future”. The detailed short analysis follows it to a reasoned decision with conditions: Short analysis “The Habitable Future” (German PDF)
First examine the present path
An impact assessment needs a reasoned description of the affected paths. The assessment guide helps build that profile.
Complete version
The longer version with application examples is available in the download section.