Defaults.Exposed

City TLDs ranked against their country codes

A city ending is sold as a civic asset. .berlin, .paris, .nyc and .london are geographic names, and under ICANN’s rules a new TLD (Top-Level Domain) that uses a city’s name for purposes tied to the city needs the support or non-objection of that city’s government. Some cities run their own. The November 2026 census grades 25 such endings, named for a city, a region, a nation or a cultural community, beside the country code each one sits inside.

On the F rate, the country code comes out ahead in 24 of the 25 pairs. The exception is .scot, at 65.4% against 65.8% for .uk. On DMARC (Domain-based Message Authentication, Reporting and Conformance) enforcement and DNSSEC (Domain Name System Security Extensions), two controls a domain owner sets at the registrar or DNS (Domain Name System) host, some endings pull ahead: seven on enforcement and seven on DNSSEC.

Each comparison sets two very different populations side by side. The civic endings hold between 5,699 and 106,139 graded names each, and their country codes between 565,337 and 12,710,322. The census records how the names in each are configured and nothing about why, so a gap describes the names under an ending and says nothing about the registry that runs it.

The 25 endings ranked

The first table ranks the endings by F rate: the share of each ending’s names carrying an F, over the names carrying a grade plus the dead ones. The host column gives the same share for the country code, and the gap is the difference in points. Every gap but .scot’s is positive.

Only one of the 25 civic endings has a lower F rate than its country code

Figure 1. The F rate of each civic ending (bar) and of its host country code (tick), over names carrying a grade plus dead names.

#endingF ratehosthost F rategap to host, pointsnames gradedplace
1.frl43.3%.nl39.7%+3.69,285Friesland
2.amsterdam45.8%.nl39.7%+6.116,259Amsterdam
3.bzh52.4%.fr49.7%+2.713,458Brittany
4.swiss58.5%.ch40.5%+18.021,167Switzerland
5.eus60.5%.es56.9%+3.614,787Basque Country
6.tirol62.8%.at52.5%+10.37,691Tyrol
7.gal63.3%.es56.9%+6.45,699Galicia
8.cat63.7%.es56.9%+6.8106,139Catalonia (Catalan)
9.scot65.4%.uk65.8%-0.412,315Scotland
10.brussels69.1%.be50.6%+18.56,434Brussels
11.paris72.7%.fr49.7%+23.018,364Paris
12.wales73.9%.uk65.8%+8.112,063Wales
13.wien74.4%.at52.5%+21.913,965Vienna
14.cymru74.6%.uk65.8%+8.88,125Wales (Welsh)
15.tokyo78.0%.jp61.0%+17.070,627Tokyo
16.nrw78.2%.de65.6%+12.616,359North Rhine-Westphalia
17.berlin78.3%.de65.6%+12.741,407Berlin
18.london78.9%.uk65.8%+13.132,937London
19.hamburg79.8%.de65.6%+14.214,349Hamburg
20.ist80.1%.tr50.8%+29.36,197Istanbul
21.nyc81.4%.us79.3%+2.156,161New York City
22.koeln84.1%.de65.6%+18.515,758Cologne
23.bayern84.2%.de65.6%+18.622,718Bavaria
24.vegas87.3%.us79.3%+8.010,559Las Vegas
25.miami88.3%.us79.3%+9.011,156Miami

The spread is wide. .frl, the Frisian ending, has the lowest F rate of the 25, 43.3%, and sits 3.6 points above .nl. .miami has the highest, 88.3%, 9.0 points above .us, which at 79.3% is the weakest host in the table. The three US city endings, .nyc, .vegas and .miami, hold three of the bottom five places. All five German endings sit above .de’s 65.6%, from .nrw at 78.2% to .bayern at 84.2%.

Taken together, the 25 endings hold 563,979 graded names, and 72.2% of them score F. Across the 526 endings the census publishes one by one, generic TLDs score 76.9% F as a group and country codes 65.3%, so the civic endings sit between the two. ICANN classes all 25 as generic TLDs.

Nearest to home and furthest from it

.scot sits 0.4 points below .uk. The next smallest gaps belong to .nyc (2.1 points above .us) and .bzh (2.7 above .fr), then .frl and .eus level at 3.6 points, above .nl and .es. The .nyc gap is small against the weakest of the eleven hosts, .us at 79.3% F. The .bzh gap is small against one of the strongest, .fr at 49.7%.

The three largest gaps are .ist at 29.3 points above .tr, .paris at 23.0 above .fr and .wien at 21.9 above .at. .bayern follows at 18.6 above .de, with .koeln and .brussels level at 18.5 points above .de and .be. Of the three largest, .ist and .paris are run by the public body itself.

The six endings run directly by a city, a national government or its agency (.amsterdam, .ist, .london, .nyc, .paris and .swiss) range from 45.8% to 81.4% F, with a median gap to their host of 15.6 points. The fourteen run by a private or non-profit registry with the government’s support have a median gap of 12.7. The five cultural endings (.bzh, .cat, .eus, .gal and .scot) have the smallest, 3.6 points. Groups of six, fourteen and five are too small to say whether who runs an ending has anything to do with its gap, and the census records nothing that would link the two.

Mail: where seven endings beat their host

The second table keeps the order of the first and shows three controls, each as a share of the names carrying a grade plus the dead ones: a DMARC record with any policy, a record that enforces with a quarantine or reject policy, and DNSSEC.

endingDMARC recordhostenforcinghostDNSSEChost
.frl47.9%58.2%29.5%31.2%34.8%56.2%
.amsterdam38.7%58.2%26.4%31.2%15.0%56.2%
.bzh31.5%39.5%11.9%13.1%29.9%22.7%
.swiss30.0%44.8%23.8%32.7%21.6%51.8%
.eus30.0%43.4%12.7%14.2%10.3%3.5%
.tirol30.6%38.3%7.3%11.7%2.8%4.3%
.gal28.6%43.4%11.2%14.2%4.0%3.5%
.cat32.4%43.4%15.3%14.2%12.4%3.5%
.scot45.9%33.9%12.8%14.9%1.6%3.7%
.brussels23.5%38.2%11.8%16.5%33.4%33.3%
.paris23.5%39.5%9.5%13.1%20.4%22.7%
.wales35.7%33.9%12.8%14.9%2.5%3.7%
.wien23.7%38.3%6.4%11.7%2.7%4.3%
.cymru31.4%33.9%14.9%14.9%2.4%3.7%
.tokyo35.7%34.9%11.7%6.9%0.3%0.7%
.nrw51.6%44.2%20.7%21.6%14.4%2.8%
.berlin55.2%44.2%27.8%21.6%1.5%2.8%
.london29.5%33.9%15.2%14.9%3.2%3.7%
.hamburg50.3%44.2%25.3%21.6%1.9%2.8%
.ist27.7%39.7%18.2%15.7%2.2%0.2%
.nyc20.0%23.7%12.2%14.7%2.1%3.0%
.koeln43.5%44.2%18.2%21.6%1.9%2.8%
.bayern37.3%44.2%17.3%21.6%1.8%2.8%
.vegas22.9%23.7%18.5%14.7%1.4%3.0%
.miami18.0%23.7%13.0%14.7%2.3%3.0%

Six of the 25 publish a DMARC record on more names than their host (.berlin, .hamburg, .nrw, .scot, .tokyo and .wales), and seven enforce it on more names (.berlin, .cat, .hamburg, .ist, .london, .tokyo and .vegas).

Three of the five German endings publish a record on more names than .de: .berlin on 55.2% against .de’s 44.2%, .nrw on 51.6% and .hamburg on 50.3%. .berlin and .hamburg also enforce above the German rate, at 27.8% and 25.3% against 21.6%, and .berlin’s lead of 6.2 points is the widest in the table. .nrw, .bayern and .koeln enforce below it, at 20.7%, 17.3% and 18.2%.

.tokyo has the next widest lead, enforcing on 11.7% of names against 6.9% for .jp, the weakest host on enforcement, 4.8 points ahead. .vegas enforces on 18.5% against 14.7% for .us, and .ist on 18.2% against .tr’s 15.7%, 2.5 points ahead. .cat and .london lead by under two points: 15.3% against .es’s 14.2%, and 15.2% against .uk’s 14.9%. The widest gaps the other way are .swiss, at 23.8% against 32.7% for .ch, the strongest host on enforcement, then .wien at 6.4% against .at’s 11.7% and .amsterdam at 26.4% against .nl’s 31.2%.

The share of names whose DMARC record names an aggregate-report address runs from 2.7% in .koeln to 24.3% in .frl, so in every ending most names publish no report address. What an enforcing record looks like, and what the report address is for, are the same under a civic ending as under any other.

DNSSEC: seven endings beat their host

Seven of the 25 have DNSSEC on more of their names than their host: .brussels, .bzh, .cat, .eus, .gal, .ist and .nrw. Four of them lead by more than six points: .nrw with 14.4% of names signed against .de’s 2.8%, the highest share of any German ending; .cat with 12.4% against .es’s 3.5%; .bzh with 29.9% against .fr’s 22.7%; and .eus with 10.3%, also against .es. The other three lead by two points or less.

Three hosts sign more than 30% of their names: .nl (56.2%), .ch (51.8%) and .be (33.3%). Under them, .amsterdam signs 15.0%, .frl 34.8% and .swiss 21.6%, while .brussels, at 33.4%, sits 0.1 points above .be and so counts among the seven.

The census records these shares and not which registrar or DNS host produced them. What signing changes and what it breaks applies under a civic ending as under the country code.

Dead names

A dead name is one where the scan got a definitive empty answer for the domain’s name servers and no address or mail record resolved, as the ghost towns piece sets out, and the census grades it F. Across the 25 endings, 1,651 of 563,979 names are dead, 0.29%, and they make up 0.4% of the Fs.

Only .swiss (1.35%) and .tokyo (1.33%) reach 1%, and twenty endings have a lower dead share than their host: .amsterdam, .bayern, .berlin, .bzh, .cat, .cymru, .eus, .frl, .gal, .hamburg, .koeln, .london, .miami, .nrw, .nyc, .paris, .tirol, .vegas, .wales and .wien. The F rates above are almost entirely live names that publish too little.

How the rates were computed

All figures come from this edition of the census. For the 25 civic endings they’re counted from the census’s per-domain records, one row per domain, and they match the per-TLD figures the census publishes. The country-code figures are those published rows.

Every share uses one denominator, the names carrying a grade plus the dead ones, which the counting page sets out for grade shares. The DMARC and DNSSEC columns use the same base, so a dead name counts as having no record.

The F rate is the share of those names graded F, and dead is the names the census marks dead. A DMARC record counts whatever its policy, including none and malformed records; enforcing means a quarantine or reject policy. DNSSEC means the zone publishes both a DNSKEY (DNS Public Key) record and a DS (Delegation Signer) record at its parent; the census checks that both are present and doesn’t validate the signatures. Gaps compare the two shares as printed, to one decimal place.

The 25 were chosen by hand from the endings the census publishes one by one, and the first table is the full list. An ending qualifies if it’s named for a city, a region, a nation or a cultural community and exists with the support of that place’s government or community body. .kiwi, .lat, .africa and .asia are in the census and left out. No public body stands behind .kiwi or .lat, and .africa and .asia cover a continent instead of a city or region. The census file carries .scot with a country label, so the grouping here comes from the hand list and from nothing in the file.

Another 30 civic endings are in the scan with fewer than 5,000 graded names, the floor the census applies before it publishes a TLD on its own. .istanbul, .ruhr, .saarland, .vlaanderen, .madrid, .barcelona, .boston, .kyoto, .osaka, .sydney and .melbourne are among them, and so is .quebec, at 4,999 names in this edition. They aren’t ranked here because their figures aren’t on the data page for you to check.

Nothing here is compared with an earlier edition, and the what we compare page says why.

Check a name under one of these endings

If you own a name under one of these endings, the free scan reads it live and shows which of the checks behind the F rate it fails. DMARC enforcement and DNSSEC are both set at the registrar or DNS host. The report for one name uses the same measurement as the tables.

Figures from the defaults.exposed census, data as of 2 November 2026. Census numbers move every month; the current values are on the census data page.