What the program forecasts about the near future — anchored on its main prediction: that automation, applied to the systems we actually run, produces more inefficiency and more work before it produces fewer jobs.
Document Status — Companion Paper (The Outlook) · Series: Informational Friction (a systems theory)
0. How to read this Outlook
These are forecasts, not results. Each is tagged [tendency] (holds under stated conditions) or [conjecture] (offered for examination, fenced), and each rests on one or more falsification tests gathered in the companion paper. The discipline is one rule: a forecast holds only while its test survives; refute the test and the forecast falls with it.
One thing governs the register of all of them. The defensible spine under the whole Outlook is gradual and lawful: systems drift toward inefficiency by their own ordinary operation, measurably and incrementally, and that drift — not dramatic collapse — is what the forecasts chiefly assert. Where collapse appears it is the rare terminal edge of the drift, fenced as conjecture.
1. The anchor forecast: inefficiency before unemployment
[tendency — falsifiable] The loudest current forecast is that AI eliminates jobs. The program predicts close to the opposite, at least first: automation, applied to the systems we actually run, raises inefficiency and the total quantity of work before it lowers the number of workers. Inefficiency rises before anything dies, and rising inefficiency demands more hands, not fewer.
The mechanism is the one the systems papers establish, not a new claim. A fully map-specified agent (Paper 2) acts only on the declared map and is blind to the off-map; inserting one does not clean the process, it deletes the integrating layer — the coherent, off-map work (positive deviation, Paper 3) that was reconciling what the map leaves disconnected — and leaves a thin conformant output that produces little. A process that produces little requires another process to complement it, which requires another to complement that. So automation, applied to a divergent estate, does not collapse the headcount; it multiplies the complementary processes needed to compensate for what the maximizer can no longer integrate. The map fills with more boxes, not fewer — friction made visible as labour.
The non-fatalist turn. Every agent who appears on the map but produces little is handed an opportunity to become a contributor — a positive deviation that absorbs the inefficiency shocks, using the automated tool as an auxiliary instrument for contribution rather than as a replacement for it. The consensus errs by confusing a role’s map appearance with its contribution (the central error the economic series names): automation removes the substitutable appearance-work and intensifies demand for the unsubstitutable contribution-work — which, the systems we build being maximizers, stays human (the coherence-capacity debt, below).
Two fences keep it honest. The boundary: on a genuinely healthy, low-gap process, automation does substitute cleanly and does cut jobs — the inefficiency-inflation claim is about divergent, high-deviance-dependence estates, so the precise statement is that the effect rises with deviance-dependence. The branch: the opening for contribution is an opportunity, not a guarantee — the coherence radar (Paper 2) predicts the system will suppress the very shift it now needs. So the outcome forks: either a healthy migration from appearance-work to contribution-work, or a swelling deviance layer that merely defers the drift — and which branch is taken depends on whether the system tolerates the coherence it now leans on more heavily. Rests on the maximizer-lethality test and the total-process-work test (companion paper).
[the precise statement] Stated tightly enough to test: in high-divergence processes, initial automation will increase exception handling, reconciliation, and complementary process work before producing net labour displacement. The test is an interaction, not a main effect, and it requires care in specification: the unit of analysis is the process, not the firm; pre-automation deviance-dependence is measured first (the off-map output fraction of Paper 3); the automation event is dated; the outcome is total process work — headcount plus complementary processes plus integration and exception-handling effort — not formal headcount alone; a matched low-divergence process serves as comparison; and the observation window is long enough to capture the complementary-process build-out. The model is
ΔWork = β1·Automation + β2·DevianceDependence + β3·(Automation × DevianceDependence),
and the program predicts β3 > 0: the more a process already ran on off-map deviance, the more automation adds total work rather than removing it. A negative or null β3 — automation reducing total work regardless of divergence — refutes the anchor.
2. The forecasts that follow
The remaining forecasts are stated briefly; each names its discipline and the test it depends on.
[tendency] 2. Self-rationalizing systems are lethal on the sick, safe on the healthy. Automated “clean-up” preserves value on a low-gap estate and destroys it on a divergent one, at a rate rising with the estate’s dependence on off-map structure — most destructive exactly where it is deployed to fix the worst mess, because the off-map survival layer is invisible to a maximizer. (Systems theory; rests on the maximizer-lethality test.)
[tendency] 3. Coherence-capacity debt rises. As estates mature and entangle and as automation spreads, the gap between the coherence a system needs and the maximization a mechanical agent can supply grows; the scarce human carriers of coherence are demanded faster than they appear. (Organizational theory; rests on the coherence-radar and maturity tests.)
[tendency] 4. The maturity inversion. More mature, more audited, more certified systems become more dependent on positive deviation, not less, and more fragile to clarity — because each control body tightens the map against the flow. (Spans all four series; rests on the maturity-inversion test — the most counterintuitive single measurement in the program.)
[tendency] 5. Deviance-dependence trends toward one. In legacy-heavy estates a growing share of real output flows off the map, and conformant execution increasingly yields zero or harm. (Systems theory; rests on the conformance-counterfactual and deviance-dependence tests.)
[tendency] 6. Discovery gaps widen. As estates grow, the divergence between the declared inventory and the traced flow grows, and the undocumented set increasingly over-represents load-bearing structure rather than junk. (Information theory / economics; rests on the discovery-gap test.)
[tendency] 7. Build-versus-buy keeps inverting; growth keeps masking hollowing. Systems increasingly acquire capability they could build, and visible growth continues to hide internal extraction — until the external-supply treadmill cannot keep up. (Economics; rests on the internalization-wedge and build-vs-buy tests.)
[tendency] 8. Clarity without revaluation accelerates the loss it means to prevent. Transformation and governance programs that resolve ownership without first recovering value are followed by faster capability loss, not slower — the unowned, which is the load-bearing thing, is what naive clarity cuts. (Spans economics and systems theory; rests on the governance-treatment comparison — the program’s decisive experiment.)
[tendency] 9. The carriers are squeezed as the system sickens. Suppression of non-optimizing-but-coherent behaviour intensifies; carriers hide or leave; and capability collapses cluster after carrier departures rather than after a measured performance shock. (Systems / organizational; rests on the coherence-radar residual and the loss-is-informational test.)
[conjecture] 10. At the economy scale, the most-revered non-tradable capabilities lose their carriers first — filled by imported labour, with abandonment intensifying where reputation is highest. Fenced, with trade and comparative advantage named as the rival on the tradable margin. (Economy conjecture; outside the firm-level core.)
[conjecture] 11. Closed systems reach a terminal crossing. A system that receives no external structural awareness can end, in the rare limit, when a load-bearing capability crosses outside the frame the system will hold — a death that clusters after that crossing, not after a real performance shock. The terminal edge of the gradual drift, the program’s hardest line to instrument. (Systems theory; rests on the F/−F regime-detection conjecture and the loss-is-informational test.)
3. The shape of the whole Outlook
Read together, the forecasts describe a near future in which the systems we run get busier and less efficient before they get smaller; in which automation, deployed as a maximizer onto divergent estates, multiplies inefficiency and destroys the off-map structure that was quietly keeping those estates alive; in which the scarce, human capacity for coherence becomes the binding constraint; and in which the one thing that would reverse the drift — structural awareness injected from outside, revaluing before it clarifies — is exactly the thing a fully map-specified agent cannot supply and a closed system cannot generate.
None of this is asserted as fact. It is the extrapolation of a falsifiable core, and it stands or falls with that core. The tests that would confirm or break each forecast are gathered, in full and with the result that would refute them, in the companion refutation paper — so that an opponent can go straight there and decide whether the program holds.
— Iván Abril Palma