engine: tune the default world to a soft program

DefaultConfig now describes a more realistic world: a moderately novel fake
(novelty 0.3), some ambient harm awareness (0.2), and a strong but imperfect
education program (programEffect 0.8) so educated students mostly, not always,
refuse. The headline shifts from 99/70/7 (a perfect program) to 100/83/21 out
of 120 (83/69/18 percent): no program >> random >> most-connected still holds,
targeting still wins by ~4x, but the program is no longer a perfect wall.
Golden values re-pinned in the engine and API tests; the preset base matches.
The forward-chance formula test now neutralises its baseline so it pins the
formula, not the tuned defaults.

internal/api/api_test.go

@@ -81,9 +81,9 @@ func TestComparisonEndpointMatchesGoldenValues(t *testing.T) {
 	// The same golden values the engine tests pin; the API must not
 	// change them in transit.
 	wantReached := map[engine.Strategy]int{
-		engine.StrategyNone:          99,
+		engine.StrategyNone:          100,
-		engine.StrategyRandom:        70,
+		engine.StrategyRandom:        83,
-		engine.StrategyMostConnected: 7,
+		engine.StrategyMostConnected: 21,
 	}
 	if len(comparison.Results) != len(wantReached) {
 		t.Fatalf("got %d results, want %d", len(comparison.Results), len(wantReached))

internal/api/scenario_test.go

@@ -29,8 +29,8 @@ func TestScenarioEndpointReturnsResultAndTopology(t *testing.T) {
 		t.Fatal(err)
 	}
 
-	if response.Result.NumReached != 7 { // the pinned golden value
+	if response.Result.NumReached != 21 { // the pinned golden value (tuned default world)
-		t.Errorf("NumReached = %d, want 7", response.Result.NumReached)
+		t.Errorf("NumReached = %d, want 21", response.Result.NumReached)
 	}
 	if response.Config != request.Config {
 		t.Errorf("config not echoed: got %+v", response.Config)

internal/engine/forwardchance_test.go

@@ -7,7 +7,13 @@ import "testing"
 // the program scales an educated student) in terms of the weight constants,
 // so tuning the weights later does not silently break the relationships.
 func TestForwardChance(t *testing.T) {
-	base := DefaultConfig() // ForwardProb 0.38, novelty/harm 0, programEffect 1
+	// A neutral baseline (the tuned DefaultConfig now carries novelty, harm
+	// awareness and a soft program); these assertions pin the formula's
+	// shape, not the default values.
+	base := DefaultConfig()
+	base.Novelty = 0
+	base.HarmAwareness = 0
+	base.ProgramEffect = 1
 
 	t.Run("default not educated is the baseline", func(t *testing.T) {
 		if got := base.ForwardChance(false); got != base.ForwardProb {

internal/engine/scenario.go

@@ -50,10 +50,10 @@ func DefaultConfig() Config {
 		EdgesPerNode:  3,
 		TriangleProb:  0.45,
 		ForwardProb:   0.38,
-		Novelty:       0, // behaviour-neutral until the model is tuned (slice 4)
+		Novelty:       0.3, // a moderately novel/shocking fake
-		HarmAwareness: 0, // "
+		HarmAwareness: 0.2, // some ambient AI-literacy / harm awareness in the year group
 		NumEducated:   36,
-		ProgramEffect: 1.0, // a perfect program: today's hard block, softened in slice 4
+		ProgramEffect: 0.8, // a strong but imperfect program: educated students mostly, not always, refuse
 		Origin:        0,
 		GraphSeed:     17,
 		ThresholdSeed: 2,

internal/engine/scenario_test.go

@@ -23,13 +23,16 @@ func runAllStrategies(t *testing.T) map[Strategy]Result {
 }
 
 func TestRunScenarioGoldenReachValues(t *testing.T) {
-	// Pinned from the first verified run (2026-06-10). The prototype's
+	// Pinned for the tuned default world (2026-06-18): a moderately novel
-	// figure showed 102/77/10 out of 120 (85%/64%/8%); ours is the same
+	// fake, some ambient harm awareness, and a strong-but-imperfect program
-	// story with different dice.
+	// (programEffect 0.8), so educated students mostly refuse rather than
+	// never forwarding. The prototype's figure showed 102/77/10 (85/64/8);
+	// the story (no program >> random >> most-connected) is the same, the
+	// dice and the soft program differ.
 	wantReached := map[Strategy]int{
-		StrategyNone:          99, // 82%
+		StrategyNone:          100, // 83%
-		StrategyRandom:        70, // 58%
+		StrategyRandom:        83,  // 69%
-		StrategyMostConnected: 7,  // 6%
+		StrategyMostConnected: 21,  // 18%
 	}
 	results := runAllStrategies(t)
 	for strategy, result := range results {

web/src/presets/deepfake-school.ts

@@ -28,9 +28,9 @@ export const deepfakeSchoolPreset: StudyPreset = {
     edgesPerNode: 3,
     triangleProb: 0.45,
     forwardProb: 0.38,
-    novelty: 0,
+    novelty: 0.3,
-    harmAwareness: 0,
+    harmAwareness: 0.2,
-    programEffect: 1,
+    programEffect: 0.8,
     numEducated: 36,
     origin: 0,
     graphSeed: 17,