There is a category of invention that is hard to recognise as an invention because it seems like a description of how things obviously are.
Alphabetical order is the clearest example. It appears to be a fact about the alphabet rather than a technique, and using it feels like following a rule rather than employing a tool.
It is a tool, it took a long time to be adopted, and it was resisted — because it does something that struck people as perverse. It arranges material in an order that has nothing to do with what the material means.
Everything that can be looked up rather than read through depends on that idea, and the consequences run substantially further than books.
Why Anybody Objected
The resistance is worth understanding because the objection was coherent.
For most of the history of written knowledge, material was organised by significance. A treatise proceeded from first principles to conclusions. A catalogue of creation began with the most important things and worked downward. A list of any kind reflected a hierarchy.
That ordering carried information. Knowing where something appeared told you something about what it was, and the structure of a work was part of its argument.
Alphabetical order destroys that completely. It places unrelated things adjacent to each other because of an accident of spelling, and it implies that the arrangement is a matter of convenience rather than of truth.
To somebody who regarded a work as an argument, that was not a neutral technical choice. It was a refusal to say what mattered.
The counter-argument is the one that eventually won: an order that carries no meaning can be used without understanding the material, which means somebody can find a thing without already knowing where it belongs.
Like our content? Follow us for more.
What an Index Actually Does
The mechanics repay a closer look because the design decisions are not obvious.
An index is a list of things a reader might look for, each paired with a location in the text — which means it is a second document, about the first, arranged on an entirely different principle.
The difficult part is deciding what goes in it. A text mentions an enormous number of things and most of them are not worth indexing, so the compiler must judge what somebody might come looking for, which requires imagining readers with purposes the author did not have.
The second difficulty is naming. A concept discussed across a book may be referred to in several ways, none of which is the term a reader would search under, so entries must be chosen as headwords rather than extracted as they appear.
And the third is structure. Entries with many locations become useless — a list of forty page numbers tells a reader nothing — so they must be subdivided, which means the compiler is imposing an analysis of the material.
That is why indexing is a skilled trade rather than a mechanical extraction, and why an automatically generated word list is not an index.
There is a point about direction worth adding. An index is read backwards relative to the text – a reader arrives knowing what they want and needing to find where it is, which is the opposite of reading, where you arrive at the position and discover what is there.
That reversal is what makes a reference work a different kind of object rather than a differently arranged one.
The Page Number Problem
A detail underlies the whole apparatus and is easy to overlook.
An index requires stable locations. Pointing at a place in a text is only useful if that place is the same in the reader’s copy as in the compiler’s.
Before printing, it was not. Every manuscript copy differed in layout, so a reference to a position in one copy was meaningless in another — which meant indexes had to reference something stable, such as a chapter division or a numbered section imposed on the text.
Printing changed that by making every copy of an edition identical, which is what made page numbers possible and therefore made the modern index possible.
That is a significant dependency. The index is frequently described as a consequence of printing, and the mechanism is specific: printing did not make indexing desirable, it made a stable address available.
And the dependency reappears whenever the address becomes unstable again, which is why a text that reflows to fit a screen cannot use page numbers and has to point at something else.
Everything Else That Uses the Idea
The principle escaped books almost immediately, and it is worth listing where it went.
Catalogues of every kind — libraries, collections, stock, membership — use it, because the person consulting a catalogue is by definition looking for something they cannot locate themselves.
Directories arrange people and businesses by name rather than by importance, which is why they can be used by strangers.
Dictionaries and encyclopaedias are the form taken to its conclusion: works with no reading order at all, designed exclusively to be consulted.
Filing systems, reference numbering and every arrangement where material must be retrieved by somebody who did not put it there use the same logic.
And the entire architecture of searching anything electronically is the same idea implemented differently: an index built separately from the material, arranged for retrieval rather than for meaning, pointing at locations.
That last point is worth sitting with. A search engine is an index in precisely the sense a book index is, solving precisely the problem, using an arrangement that carries no meaning so that it can be used without understanding.
What Is Lost
The trade-off is real and is worth stating, because the technique is not free.
An index removes context. A reader who arrives at a page through it encounters material without the argument that led to it, which is fine for a fact and misleading for a claim that depended on what preceded it.
It encourages extraction over reading, which is frequently what is wanted and occasionally what is not.
It imposes the compiler’s analysis. What can be found is limited by what somebody decided to index, and material that was not anticipated is effectively invisible even though it is present.
And it makes works usable without being read, which changed how texts were engaged with in ways that were noticed and complained about at the time.
Those objections are the same ones made about every retrieval technology since, in almost the same words, which suggests they are objections to the capability rather than to any particular implementation.
Ordering Things That Do Not Have Letters
A complication worth including is that alphabetical order only works on words, and a great deal of material is not words.
Numbers have their own order, which is unambiguous until you have to interleave them with words, at which point somebody must decide whether a numeral sorts before or after a letter, and the choice is arbitrary.
Names introduce the problem of which part to sort on, since a person’s name has components of different kinds and conventions differ about which is the entry point — and the answer varies between traditions in ways that cannot all be accommodated by one rule.
Titles beginning with articles present the same difficulty, which is why such words are conventionally skipped for sorting purposes and why that convention has to be stated rather than being obvious.
Material in different scripts cannot be interleaved at all without a decision about how to map between systems, which is a substantial problem in any collection drawing on more than one.
And things that are not named at all — images, objects, sounds, places — must be given a verbal handle before they can be arranged alphabetically, which means somebody is describing them, and the description determines whether they can be found.
That last point is the persistent one. Anything retrievable by name is retrievable only through the name somebody assigned it, which means the ordering is neutral and the naming is not — and the naming is where every difficulty in cataloguing has always been.
Why It Deserves More Credit
The reason this is worth thinking about is what it demonstrates about tools that disappear.
Alphabetical arrangement is invisible as a technique because it has become the default, and defaults do not look like choices. Nobody encountering an index thinks about the decision to abandon meaningful order, because there is no visible alternative to compare it with.
But it was a decision, it was argued about, and it produced a capability that had not previously existed: the ability to find something in a body of material without any prior knowledge of how that material is arranged.
That is the whole of what a reference work does, it is the whole of what searching does, and it rests on somebody being willing to arrange things in an order that means nothing at all.
Which is a reasonable thing to remember while looking something up. The arrangement that makes it possible was once regarded as an abdication, and its uselessness as a statement about the world is exactly the property that makes it useful to everybody.
Which is the odd thing about good tools. The ones that work best frequently do so by refusing to be clever, and an ordering that declines to say anything about the world turned out to be the one that could be used by anybody in it.
Like our content? Follow us for more.
