How the canon was built

How the 21st-century canon was built

A note on method for The 250 Best Electronic Albums of the 21st Century (2000–2025).

This ranks albums as albums. Not favourite artists, not biggest sellers, not the best DJ tools. It was built the way an editorial panel works, drawing on the critical traditions of Resident Advisor, The Wire, Pitchfork, FACT, Mixmag, DJ Mag and the specialist dance press: broad expertise, argued out record against record, never one critic's taste.

Dance-first, instrumental, production-led

Dancefloor greatness and field-defining artistry count equally. Plenty of records here carry ambient, pop or jazz among their tags, and the tags describe ingredients. The scope line is about what a record is for. Albums whose whole mode is song-led pop, rap or beatless ambience were held to a stricter bar: they chart only when the production, not the voice or the stillness, is the engine. Most were kept out. Tim Hecker's Ravedeath, 1972 is the clearest case of one that earned its way in.

That rule costs the list some obvious records and it is the rule anyway. A canon that admits every good album with a synthesizer on it is a genre survey.

Two tiebreaks

Most records separate themselves. When two do not, there are two tiebreaks.

A dancefloor record beats a non-floor record. An album built to work in a room outranks one built to be admired in it.

An underground record beats a mainstream one. If a club record and a pop record are otherwise level, the club record goes higher.

These are tiebreaks, not scoring categories. They matter only when the records are otherwise level.

What earns a place

Every album was judged on seven things: artistic achievement, album coherence, innovation, demonstrated influence, durability, scene significance, and emotional or conceptual force. Those judgements band a record into a merit tier, S down to D. Inside a band, records are ordered head-to-head, so no two share a rank by accident.

Influence and durability take time to prove. Anything from roughly 2018 on is judged on the impact it has already had, not on the canon it might one day join.

Merit only, no quotas

Nothing is credited for where it came from or which scene it fills. Any record that survived only on geography, scene breadth or artist pedigree was cut.

Bias diagnostics were run, and run explicitly: era balance, genre prestige, Anglo-American and Western European dominance, and both popularity and obscurity bias.

Electronic-music lists have a habit of rewarding records you sit with over records that actually work in a club. This one pushes back. House, techno, garage, jungle, bass and global club music are not treated as a lower form of the art because people dance to them.

The list leans Anglo-American and Western European. Those histories have been written down more completely than Latin American, African, Japanese and other regional club traditions, and there is more of all four than this list has been able to verify. The imbalance is real and readers should know it is there.

A diagnostic is an instruction to look harder at work that may have been under-valued. It is never a licence to lower the bar.

Honesty about precision

Adjacent ranks are not claims. Tiers are. The order is most reliable at the very top. Through the middle tiers, treat ten places either way as drawing order, and fifteen at the tail. A one-point score gap is never an argument.

If a site is going to number 250 records, it should also say how seriously to take the numbers. That is the honest range.

How it was verified

The list went through eligibility and scope screening, a full dance-first re-scoring, a research pass to correct recall bias (records under-credited only because they are less famous), a complete head-to-head re-ranking of every tier, a five-pass merit audit, and a final placement audit.

Every ruling was written down as it was made. What came out is one ordered canon, argued and defensible from #1 to #250.

How the 20th-century canon was built

A note on method for The 250 Best Electronic Albums of the 20th Century (1975–1999).

The list has a working name: The Rave Continuum. Two hundred and fifty records, ordered on merit, each carrying the argument for why it sits where it does.

Merit sets the order. Three other measures argue with it.

The rank you see is merit. Every album also carries three placings that are meant to disagree with it.

Dancefloor. Where the record ranks by how it works on a floor.

Influence. Where it ranks by what it changed.

External poll. Where the record placed in outside critical polling of the same era. A dash means it didn't place. It is a reference point, not an input: it records where a record landed elsewhere, and it did nothing to decide where it lands here.

The gaps do the work. Selected Ambient Works 85-92 is first on merit and 193rd on the floor. Eighty-eight of the 250 sit more than a hundred places from their own dancefloor rank. Read the three against the main order and you learn what a record is actually for, which one number can never tell you.

A third of the list never placed in that outside polling at all. Those records are not lesser for it. They are the ones this canon is making a case for.

Where the years stop, and the one place they bend

The canon runs 1975 to 1999. The year 2000 belongs to the 21st-century list, and nothing crosses that line in either direction.

One record sits outside it. Tangerine Dream's Phaedra is from 1974, and it is here at 129 because it is the album where the sequencer stops being an effect and becomes the instrument. A canon built on that idea cannot begin after the record that proves it.

The seam with the 21st century

The two lists share no albums. A placing here confers nothing there. An artist ranked in this canon starts level in the other one.

How the genre lists were built

A note on method for the twenty ranked genre lists.

Twenty traditions, ten records each, with the ten to twenty band underneath and the near misses named. They are shorter than the canons and they are not extracts from them. Each was built from its own field and ranked on its own terms, which is why a record can top a genre list and sit halfway down a canon, or chart on one and appear nowhere on the other.

The scope line is different, and that is the point

The canons are dance-first. Song-led pop, rap and beatless ambience were held to a stricter bar there, and most of it was kept out. A genre list has no such rule, because the genre supplies the scope instead.

That is why Depeche Mode, OMD and the Human League are ranked on the synth-pop page and absent from both canons. The canons ask what a record is for. A genre list asks whether it is one of the best records that tradition produced, and pop is what synth-pop is.

Each page states its own boundary in the first paragraph, and several of them state it at length, because the boundary is usually the argument. Where breakcore stops and gabber starts, whether kosmische means a lineage or a set of instruments, whether trip-hop is defined by a voice: those are the decisions the lists were built on, and they are written down rather than assumed.

Overlap is expected

The two canons share no albums. The genre lists share plenty, with the canons and with each other.

That is deliberate. House and deep house are a parent and a specialist, and both should hold the records that define them, so the same album tops both. Techno and Detroit techno overlap by design. IDM's number one is ambient's number two. Sixteen records appear in more than one top ten, and where two lists rank the same album they argue for it differently, because they are asking different questions about it.

A record's rank on one page confers nothing on any other.

Two exceptions to what counts as an album

Some traditions never pressed albums. Basic Channel released twelve-inches for two years and the CD that gathers them is the only form in which that music exists as something you can sit down with. A rule written to keep out DJ mixes would delete the entire catalogue.

So a collection is eligible when it gathers one artist's own material, that material had no earlier album release, it was issued with the artist's involvement as the canonical document, and it is sequenced as a listening object rather than a chronology. Every record admitted that way is flagged where it is ranked. Fifteen of them are.

The second exception is length. A single twelve-inch of original material counts as an album where it runs to album length in its era's terms and was received as one. Robert Hood's Minimal Nation is tagged an EP on its 1994 pressing and an album on most of what followed, runs to album length, and has been treated as one for thirty years.

Neither exception admits a DJ mix, a various-artists compilation or a hits package, and neither is a licence to rank a record because its maker is important. Loefah has no album and is not ranked anywhere.

Ten is a target, not an entitlement

A list is allowed to be shorter than ten if the field will not carry ten. None currently is, but the rule is what stops a page padding itself.

Every page names the size of the field it chose from, and every page names the records that nearly made it and why they did not. Those two things are mandatory rather than decorative. A ranking without them is an assertion; with them it is an argument someone can disagree with.

The shape of the result

Fifty-six per cent of the two hundred records in the tens come from the 1990s, and the median year is 1997.

Some of that is the music. Trip-hop lasted about a decade, jungle's album era ran six years, and the album was briefly the natural unit for electronic music in a way it had not been before and has not been since. Kosmische is nine-tenths 1970s and deconstructed club is entirely 2010s, which is what the same method produces when a tradition sits elsewhere.

Some of it may be us. Influence is easier to see at distance, and a record from 1996 has had thirty years to be argued with while one from 2021 has had four. The pools were widened deliberately to test that, and the number barely moved. Either the additions were not good enough or they were judged by someone who knows the older material better, and both are possible.

The figure is here so the lists can be read with it in mind.