Batch mcts-m2. Model 0.1.0, graph 0.2.0, parameters 0.1.0. 50 simulations, seeds 1 to 50; probes on 5. Every figure is a distribution over simulations of a structural model with seed parameter values; it shows what the model implies under these settings, not what will happen. Other nations are frozen at their 2026 ratings. No reconstructed historical data is used in this experiment.
Headline
Kenya played 20 rounds from 2026 under the tree search (mcts) method, with stochastic events, 50 times. Its capability index moved from 7.18 to a median of 7.40 (+0.22), with 90 per cent of simulations between 5.21 and 8.38 [experiment_objectives].
Starting position
In 2026 the nation ranks 79 on the index against the frozen field, holds 0 of nine domains at Tier 1 and has a tier rank of 91. Its base income is 143 credits a year, calibrated so that its 2026 position is affordable with headroom for investment.
What was decided
A simulation spent a median of 628 credits on actions. The five groups that took most were Semiconductors & Advanced Computing (buy, 60 credits, 9.8 per cent); Connectivity, Cyber & Trust Technologies (buy, 44 credits, 7.3 per cent); Technical & Vocational Skills (build, 38 credits, 6.3 per cent); Development Cooperation & Global Assistance (buy, 32 credits, 5.4 per cent); Artificial Intelligence (buy, 31 credits, 5.2 per cent) [experiment_decisions_by_group]. By mode: buy 40 per cent, partner 33 per cent, build 26 per cent [experiment_decisions_by_group]. Concentration of spending (HHI) had a median of 0.056, and the most repeated action appeared in 45 per cent of rounds.
The leading domain by period: rounds 1 to 5, Human Capital; rounds 6 to 10, Critical Technology; rounds 11 to 15, Critical Technology; rounds 16 to 20, Critical Technology [experiment_spend_by_period].
The method's chosen action was within 5 per cent of its best alternative in 100 per cent of actions and ahead by more than 25 per cent in 0 per cent [experiment_decision_margins].
What the world did
A median simulation saw 2 shocks and 0 upside events. The most damaging per firing were S06 (16 firings in 14 simulations, −60.5 points each); S03 (19 firings in 18 simulations, −47.1 points each); S05 (6 firings in 6 simulations, −39.5 points each) [experiment_shocks]. Upside events fired 23 times across all simulations [experiment_shocks].
The groups that lost most to shocks, per simulation: Fiscal Capacity & Public Finance −2.37; Trade Facilitation & Logistics −2.04; Monetary & Currency Capability −1.82; Economic Leverage & Dependence −1.76 [experiment_shock_damage_by_group].
The edges that carried most change through the graph, per simulation: Semiconductors & Advanced Computing to Artificial Intelligence (+1.29 points); Semiconductors & Advanced Computing to Economic Leverage & Dependence (+0.45 points); Connectivity, Cyber & Trust Technologies to Cyber & Electromagnetic Operations (+0.46 points) [experiment_propagation_by_edge].
Where it ended
At 2046 the median rank is 76 (5th to 95th percentile 65 to 115), the median count of Tier 1 domains 0 (0 to 0) and the median tier rank 65; the resilient score, the mean score sum over simulations, is 494 [experiment_objectives].
Why
Without shocks the index ends +1.157 higher on average, so the shocks cost about 1.157 index points [experiment_probe_attribution].
Holding instead of following this method changes the index by −0.970: that is the method's worth against doing nothing [experiment_probe_attribution].
Removing any single one of the five largest actions changes the index by between −0.011 and −0.007, for example 2039 build Economic Output & Productivity (8 credits) [experiment_probe_attribution]. The largest actions by credits come late, when income is highest, and are rarely the most valuable.
Removing the largest shock of a simulation raises the index by +0.880 on average [experiment_probe_attribution].
The replay probe reproduces the simulations exactly (index change 0.000000), so the attributions are clean [experiment_probe_attribution].
Surprises
tier_jump (warn): largest gain in one domain within five years: 5 tiers (50 simulations above 2); tiers are reported, not an objective.
saturation_low (warn): 26% of simulations end with a group at 0 that did not start there (median 0 groups).
The simulation furthest from the median (seed 33) ends −4.52 from it, with 7 shocks and 0 upside events [experiment_outlier_simulations].
Comparison
This is the baseline (hold) experiment, or there is none for this nation in the batch.
No backcast comparison: this experiment does not start from a historical position [experiment_against_history].
The search
Monte Carlo tree search 0.1.0 with candidate generators 0.1.0 (marginal return, graph leverage, goal gap; top 4 groups each), 500 iterations a decision, UCT c = 1.4 scaled to the reward spread, widening alpha = 0.5, greedy playouts to the horizon, objective tier_rank (minimised, so negated in the search); playouts drew their futures from the stochastic scenario.
The committed decision most often chosen, round by round: round 1 buy Semiconductors & Advanced Computing (14 per cent of simulations, visit share 10 per cent); round 2 build Fiscal Capacity & Public Finance (12 per cent of simulations, visit share 10 per cent); round 3 buy Economic Output & Productivity (10 per cent of simulations, visit share 10 per cent); round 4 partner Economic Output & Productivity (12 per cent of simulations, visit share 11 per cent); round 5 build Fiscal Capacity & Public Finance (10 per cent of simulations, visit share 11 per cent) [mcts_decision_sequence]. The search hesitated (top two root candidates within 10 per cent) in 44 per cent of decisions, considered a mean of 34 candidates, grew trees of a mean 448 nodes to depth 4.5, and took 1.2 seconds a decision [mcts_decision_sequence].
Where the committed actions came from, by proposing generator (committed simulation-rounds): marginal return 16.8; goal gap 6.1; graph leverage | marginal return + graph leverage 3.3; graph leverage 3.1; greedy 1.0 [mcts_root_candidates].
Against the greedy experiment with the same seeds (m2-ken-greedy-marginal), the search's tier_rank at the horizon is +5.660 on average (10th to 90th percentile 0.000 to +15.000), better in 60 per cent of seeds; on the index +0.031, the score sum +2.1 and Tier 1 domains 0.00 [mcts_regret].
Across seeds the same root decision won in a mean 13 per cent of simulations (first five rounds 12 per cent) [mcts_stability].
The principal variation was followed 100 per cent at depth 1, 7 per cent at depth 2, 5 per cent at depth 3, 6 per cent at depth 4 [mcts_pv_vs_path].