Two Four-F systems
The site uses two different Four-F frames. They should not be collapsed.
| Frame | Terms | Role |
|---|---|---|
| Operational Four Fs | Fast, Flexible, Frugal, Federated | Design constraints for the system |
| Evolutionary Four Fs | Feed, Fork, Fight, Flee | Lifecycle operations inside the ecology |
The operational frame defines what kind of system should be built. The evolutionary frame defines how it adapts.
Cross-mapping
| Evolutionary operation | Fast | Flexible | Frugal | Federated |
|---|---|---|---|---|
| Feed | collect low-latency traces | accept multiple data and task types | store only useful evidence | collect local feedback without centralizing raw data |
| Fork | generate small candidates quickly | try multiple operators | respect mutation budgets | create site-specific descendants |
| Fight | benchmark under latency limits | compare interchangeable modules | include cost in fitness | evaluate across heterogeneous clients |
| Flee | route around failures immediately | swap modules cleanly | retire expensive dead weight | isolate bad local updates |
Teleodynamic coupling
Teleodynamic control couples these frames through the resource ledger. A fast system that is not frugal may collapse under cost. A flexible system that is not governed may become unsafe. A federated system without Fight and Flee may spread weak or malicious updates.
FUNCTION four_f_cycle(event, ecology, policy)
fed <- FEED(event, policy.data_rules)
candidates <- FORK(fed, policy.allowed_operators)
results <- FIGHT(candidates, policy.evaluation_suites)
ecology <- FLEE_OR_PROMOTE(ecology, results, policy.viability_rules)
RETURN ecology
END FUNCTIONDesign rule
Every architectural feature should support at least one operational F and one evolutionary F. For example, a model registry supports Flexible replacement and Fight evidence. A rollback pipeline supports Fast recovery and Flee. A local telemetry store supports Federated learning and Feed.
Anti-patterns
- Fast but not Frugal: low latency only by overprovisioning wastefully.
- Flexible but not Governed: anything can be swapped, including unsafe candidates.
- Federated but not Observable: local variation occurs without comparable evidence.
- Fork without Flee: the population grows forever.
- Fight without Feed: tests become stale.
- Feed without Fight: telemetry accumulates but selection never improves.
Source reports used for this guide
These reports are preserved verbatim in the site archive. The guide above is an editorial synthesis and may narrow, qualify, or reorganize claims from the source material.