History

The run fills the grid a unit at a time — one capability group at one snapshot — and everything above is drawn from what exists, never from a model of what does not. Two things are worth reading before the chart is taken at face value.

Measure	Value	What it means
Cells coded	143,077	of 143,080 in the sovereign grid
Units adjudicated	1,460	of 1,460 group-snapshots
Units adjudicated by 1 seat	1	a single model’s judgement, with no panel to adjudicate against — the weakest cells in the series
Units adjudicated by 2 seats	1,459	2 independent models, mean of their integers
Mean panel SD	±0.52	across the units that had more than one seat

The panel ran a seat short. It is designed as three vendors across three jurisdictions — China, France and the United States. The Mistral seat was rate-limited by its own account within minutes of the run starting and never recovered enough to adjudicate, so the scores here are the work of two vendors in two jurisdictions, DeepSeek and OpenAI. Cross-vendor bias is the panel's main purpose and two seats measure it less well than three; the spread above is between two models, not three.

And the usual caveat, louder. These are candidate ratings from language models with no source documents attached, now extended over five centuries. The 2026 run records that as an open problem for the present day; before about 1800 the record the models are drawing on thins badly. The early part of the chart is the part to trust least.

The grid

How many of the 197 territories were sovereign states at each snapshot, from the lineage data. This is the shape of the run: a territory under someone else's government is not rated at that snapshot, because a national capability score for it is either a different question or void. Half the full 197 × 20 grid falls away on this rule.

The baseline fed to the grid

Every nation carries its published 2026 capability-group score into the prompt as the last point of the series. It anchors the scale; it is explicitly not a prior for the historical value, and the protocol says so in as many words — a nation at 17 today was very often at 0 in 1800. The domain means below are where the 2020 end of every series sits.

Domain	Groups	2026 world mean	Range	Leader at the endpoint
Governance & Integrity	8	8.7	1.5–17.4	🇫🇮 Finland
Human Capital	9	8.5	1.2–17.2	🇸🇬 Singapore
Financial Strength	6	8.1	1.1–18.1	🇺🇸 United States
Strategic Infrastructure	10	7.8	1.2–18.4	🇺🇸 United States
Trade & Investment	7	7.7	0.5–18.1	🇸🇬 Singapore
Production & Innovation	7	7.3	0.2–18.3	🇺🇸 United States
Information & Influence	6	6.9	1.7–18.1	🇺🇸 United States
National Security	10	4.9	0.3–19.8	🇺🇸 United States
Critical Technology	10	3.5	0.0–19.8	🇺🇸 United States

The series

The world mean on each dimension at each snapshot, across the 196 nations coded so far. A line breaks where the run has not reached rather than bridging the gap.

Nations with a series

#	Nation	Snapshots coded	Earliest	2026
1	🇺🇸United States	17	1800	18.0
2	🇫🇷France	20	1500	16.6
3	🇬🇧United Kingdom	17	1800	16.5
4	🇩🇪Germany	18	1700	16.4
5	🇯🇵Japan	20	1500	16.3
6	🇰🇷South Korea	8	1950	16.0
7	🇨🇳China	20	1500	15.6
8	🇳🇱Netherlands	19	1600	15.5
9	🇸🇪Sweden	19	1600	15.3
10	🇸🇬Singapore	6	1970	15.2
11	🇨🇦Canada	14	1875	15.1
12	🇦🇺Australia	13	1900	14.9
13	🇨🇭Switzerland	20	1500	14.7
14	🇳🇴Norway	12	1910	14.6
15	🇫🇮Finland	11	1920	14.5
16	🇮🇱Israel	8	1950	14.5
17	🇹🇼Taiwan	8	1950	14.0
18	🇩🇰Denmark	20	1500	13.9
19	🇪🇸Spain	20	1500	13.8
20	🇮🇹Italy	20	1500	13.7
21	🇧🇪Belgium	16	1825	13.5
22	🇳🇿New Zealand	12	1910	12.8
23	🇦🇹Austria	20	1500	12.8
24	🇮🇳India	8	1950	12.6
25	🇵🇱Poland	14	1500	12.4
26	🇷🇺Russia	20	1500	12.3
27	🇹🇷Türkiye	20	1500	12.2
28	🇮🇪Ireland	11	1920	12.1
29	🇦🇪United Arab Emirates	6	1970	11.9
30	🇧🇷Brazil	16	1825	11.7
31	🇵🇹Portugal	20	1500	11.7
32	🇨🇿Czechia	11	1920	11.5
33	🇪🇪Estonia	6	1920	11.5
34	🇲🇾Malaysia	7	1960	11.5
35	🇸🇦Saudi Arabia	10	1930	11.2
36	🇲🇽Mexico	16	1825	10.8
37	🇱🇺Luxembourg	14	1875	10.8
38	🇿🇦South Africa	12	1910	10.8
39	🇱🇹Lithuania	6	1920	10.7
40	🇹🇭Thailand	20	1500	10.7
41	🇨🇱Chile	16	1825	10.6
42	🇬🇷Greece	16	1825	10.6
43	🇭🇺Hungary	12	1500	10.5
44	🇸🇮Slovenia	4	1990	10.5
45	🇷🇴Romania	14	1875	10.2
46	🇸🇰Slovakia	4	1990	10.1
47	🇮🇩Indonesia	8	1950	10.0
48	🇱🇻Latvia	6	1920	9.9
49	🇻🇳Vietnam	15	1500	9.7
50	🇮🇸Iceland	9	1940	9.6
51	🇭🇷Croatia	4	1990	9.5
52	🇺🇦Ukraine	4	1990	9.4
53	🇶🇦Qatar	6	1970	9.4
54	🇲🇦Morocco	14	1500	9.4
55	🇦🇷Argentina	16	1825	9.2
56	🇨🇴Colombia	16	1825	9.1
57	🇪🇬Egypt	12	1825	9.0
58	🇷🇸Serbia	14	1875	9.0
59	🇧🇬Bulgaria	14	1875	9.0
60	🇲🇹Malta	7	1960	8.9
61	🇵🇭Philippines	8	1950	8.8
62	🇵🇪Peru	16	1825	8.6
63	🇺🇾Uruguay	16	1825	8.5
64	🇨🇾Cyprus	7	1960	8.5
65	🇮🇷Iran	19	1600	8.2
66	🇰🇿Kazakhstan	7	1500	8.1
67	🇲🇺Mauritius	6	1970	8.1
68	🇴🇲Oman	18	1700	8.0
69	🇵🇰Pakistan	8	1950	8.0
70	🇨🇷Costa Rica	15	1850	7.8
71	🇹🇳Tunisia	13	1500	7.7
72	🇬🇪Georgia	7	1500	7.5
73	🇰🇼Kuwait	7	1960	7.5
74	🇯🇴Jordan	8	1950	7.5
75	🇳🇬Nigeria	7	1960	7.4
76	🇧🇭Bahrain	6	1970	7.3
77	🇱🇰Sri Lanka	12	1500	7.3
78	🇱🇮Liechtenstein	17	1800	7.2
79	🇰🇪Kenya	7	1960	7.2
80	🇵🇦Panama	13	1900	7.2
81	🇬🇭Ghana	7	1960	7.0
82	🇩🇿Algeria	11	1500	7.0
83	🇧🇾Belarus	4	1990	6.9
84	🇩🇴Dominican Republic	15	1850	6.8
85	🇦🇿Azerbaijan	4	1990	6.8
86	🇪🇨Ecuador	16	1825	6.8
87	🇧🇩Bangladesh	6	1970	6.7
88	🇹🇹Trinidad & Tobago	7	1960	6.7
89	🇦🇲Armenia	4	1990	6.7
90	🇲🇰North Macedonia	4	1990	6.6
91	🇸🇳Senegal	7	1960	6.4
92	🇲🇪Montenegro	5	1875	6.3
93	🇲🇩Moldova	4	1990	6.3
94	🇯🇲Jamaica	7	1960	6.2
95	🇦🇱Albania	12	1910	6.1
96	🇧🇼Botswana	6	1970	6.1
97	🇳🇦Namibia	4	1990	6.0
98	🇨🇺Cuba	13	1900	5.9
99	🇷🇼Rwanda	7	1960	5.9
100	🇺🇿Uzbekistan	10	1500	5.9
101	🇲🇨Monaco	20	1500	5.9
102	🇹🇿Tanzania	7	1960	5.8
103	🇧🇦Bosnia & Herzegovina	4	1990	5.8
104	🇵🇾Paraguay	17	1800	5.8
105	🇲🇳Mongolia	13	1500	5.7
106	🇪🇹Ethiopia	18	1500	5.7
107	🇸🇻El Salvador	15	1850	5.7
108	🇬🇹Guatemala	15	1850	5.7
109	🇧🇳Brunei	12	1500	5.6
110	🇧🇧Barbados	6	1970	5.6
111	🇺🇬Uganda	7	1960	5.5
112	🇨🇮Côte d'Ivoire	7	1960	5.4
113	🇸🇨Seychelles	5	1980	5.3
114	🇧🇸Bahamas	6	1970	5.2
115	🇫🇯Fiji	6	1970	5.2
116	🇦🇩Andorra	20	1500	5.2
117	🇧🇴Bolivia	16	1825	5.1
118	🇭🇳Honduras	15	1850	5.0
119	🇸🇲San Marino	20	1500	5.0
120	🇳🇵Nepal	17	1800	5.0
121	🇿🇲Zambia	7	1960	4.9
122	🇰🇭Cambodia	11	1500	4.9
123	🇰🇬Kyrgyzstan	4	1990	4.8
124	🇨🇻Cabo Verde	5	1980	4.7
125	🇽🇰Kosovo	2	2010	4.7
126	🇮🇶Iraq	10	1930	4.7
127	🇱🇧Lebanon	9	1940	4.7
128	🇲🇻Maldives	14	1500	4.6
129	🇦🇴Angola	5	1980	4.6
130	🇻🇪Venezuela	16	1825	4.6
131	🇬🇾Guyana	6	1970	4.5
132	🇨🇲Cameroon	7	1960	4.4
133	🇳🇮Nicaragua	15	1850	4.4
134	🇦🇬Antigua & Barbuda	5	1980	4.4
135	🇿🇼Zimbabwe	6	1970	4.3
136	🇰🇵North Korea	8	1950	4.3
137	🇲🇿Mozambique	6	1970	4.1
138	🇧🇯Benin	7	1960	4.1
139	🇹🇬Togo	7	1960	4.1
140	🇬🇩Grenada	6	1970	4.0
141	🇸🇷Suriname	5	1980	4.0
142	🇧🇿Belize	5	1980	4.0
143	🇹🇯Tajikistan	4	1990	4.0
144	🇬🇦Gabon	7	1960	3.9
145	🇧🇹Bhutan	14	1700	3.9
146	🇲🇬Madagascar	11	1800	3.9
147	🇲🇲Myanmar	14	1500	3.9
148	🇩🇲Dominica	5	1980	3.9
149	🇱🇨Saint Lucia	5	1980	3.9
150	🇰🇳Saint Kitts & Nevis	5	1980	3.8
151	🇵🇬Papua New Guinea	5	1980	3.8
152	🇲🇼Malawi	7	1960	3.8
153	🇻🇨Saint Vincent & Grenadines	5	1980	3.8
154	🇸🇿Eswatini	6	1970	3.7
155	🇱🇦Laos	11	1500	3.7
156	🇸🇱Sierra Leone	7	1960	3.6
157	🇲🇷Mauritania	7	1960	3.6
158	🇹🇲Turkmenistan	4	1990	3.6
159	🇼🇸Samoa	7	1960	3.5
160	🇧🇫Burkina Faso	7	1960	3.5
161	🇱🇸Lesotho	6	1970	3.4
162	🇲🇱Mali	7	1960	3.4
163	🇬🇲Gambia	7	1960	3.3
164	🇹🇴Tonga	6	1970	3.2
165	🇩🇯Djibouti	5	1980	3.2
166	🇨🇬Congo	7	1960	3.2
167	🇬🇳Guinea	7	1960	3.1
168	🇳🇪Niger	7	1960	3.1
169	🇱🇾Libya	10	1700	3.0
170	🇹🇱Timor-Leste	3	2000	3.0
171	🇱🇷Liberia	15	1850	2.9
172	🇻🇺Vanuatu	5	1980	2.9
173	🇸🇧Solomon Islands	5	1980	2.9
174	🇨🇩DR Congo	7	1960	2.9
175	🇸🇹São Tomé & Príncipe	5	1980	2.9
176	🇵🇼Palau	4	1990	2.8
177	🇬🇶Equatorial Guinea	6	1970	2.8
178	🇸🇩Sudan	10	1600	2.6
179	🇧🇮Burundi	7	1960	2.5
180	🇲🇭Marshall Islands	4	1990	2.5
181	🇫🇲Micronesia	4	1990	2.3
182	🇰🇲Comoros	5	1980	2.3
183	🇹🇩Chad	7	1960	2.3
184	🇸🇾Syria	8	1950	2.2
185	🇰🇮Kiribati	5	1980	2.2
186	🇹🇻Tuvalu	5	1980	2.1
187	🇬🇼Guinea-Bissau	6	1970	2.1
188	🇭🇹Haiti	16	1800	2.0
189	🇪🇷Eritrea	4	1990	2.0
190	🇳🇷Nauru	6	1970	2.0
191	🇻🇦Holy See	16	1500	1.8
192	🇨🇫Central African Rep.	7	1960	1.7
193	🇾🇪Yemen	15	1700	1.6
194	🇸🇴Somalia	7	1960	1.6
195	🇦🇫Afghanistan	15	1800	1.5
196	🇸🇸South Sudan	2	2010	1.4

What this run is, and what it cannot be

One ladder, anchored at the present. The scale does not move with the century: a railway network in 1850 is judged against the same 0–20 ladder as transport capability in 2026, not against 1850. That is what makes the series comparable across five centuries, and it means most nations sit in Inception or Nascent on most groups before about 1900. An index where every century looks the same would be the broken one.

The baseline anchors and may also bias. Feeding the 2026 score in keeps the scale consistent, and it gives a coder something to compress toward. The protocol forbids interpolating and says so explicitly, but the honest reliability check is the round-1 spread between coders who each saw the same anchor, exactly as in the 2026 run.

No sources, five centuries. The 2026 run already records “no source documents” as an open problem. Extending to 1500 makes that the dominant weakness, not the API bill: these would be candidate ratings produced from what the models carry, over a period where the record thins the further back it goes. Treat anything before 1800 as the weakest part of the series.

The method, the store and the runner are in history/, separate from the 2026 project in analysis/; neither writes to the other's data. SeeAnalysis for the present-day method this one inherits.

Where each nation ranked, 1500–2020

How complete this is

The grid

The baseline fed to the grid

The series

Nations with a series

What this run is, and what it cannot be