This final article in the series looks at concrete web directories — real examples, drawn from the form’s history and its present — and at the dimensions along which they differ from one another. It is worth saying at the outset what the article deliberately does not do: it does not rank the directories it discusses, and the reason is not caution but accuracy. A ranking implies a single scale of merit, and there is no single scale on which a scholarly subject gateway and a local business directory can be meaningfully placed, because they are built for different people doing different things. The honest comparison is a map rather than a league table, and the aim of this article is to give the reader the map — the set of dimensions that genuinely distinguish one directory from another — so that the right comparison can be made for any particular need.
As in the earlier articles, claims drawn from peer-reviewed research are cited by author and year and listed at the end; observations about the present state of the field rest on industry reporting and are identified where they occur.
Why directories resist a single ranking
The instinct to rank is strong, and for some things it is reasonable; but it misfires when applied across directories, and the misfire is worth understanding before the examples are introduced. To rank is to assert that the things ranked differ in degree along one shared dimension — that one is more of the same good thing than another. Directories do not differ that way. They differ in kind. A subject gateway built to give researchers a vetted set of scholarly sources and a local directory built to help a resident find a plumber are not a better and a worse version of one underlying object; they answer unrelated needs, and a measure that rated one above the other would simply be smuggling in a preference for one need over the other.
There is a further reason for restraint, and it is not merely a matter of logic. Espeland and Sauder (2007), in their study of how public rankings affect the things they rank, showed that rankings are reactive: the act of measuring and ordering changes the behaviour of the ranked, often pushing them to optimize for the measure rather than for the underlying quality the measure was meant to capture. A published ranking of directories would, in time, do the same — it would reward whatever the ranking happened to count and quietly penalize the directories that were good in ways it did not. The first article of this series described one episode of exactly this dynamic, when directories reshaped themselves to serve a search engine’s ranking signals rather than their readers. The sounder approach, and the one this article takes, is to lay out the dimensions of difference plainly and let a reader judge fitness for a purpose, because fitness for a purpose is the only assessment that a directory can be fairly held to.
The dimensions along which directories differ
Set a handful of real directories side by side and the differences sort themselves into a small number of dimensions. The most visible is scope, or breadth: a directory may be general, attempting every subject, or niche, covering one field in depth, with a wide middle ground between. Closely related but distinct is focus — the kind of thing the directory is organized around. Some directories are organized by subject, as the previous article described; others are organized by place, restricting themselves to a city or region; others are organized around commerce, cataloguing organizations and the services they sell. A directory’s scope and its focus together fix what territory it claims.
A second cluster of dimensions concerns how a directory is populated and by whom. The basis of inclusion varies: a directory may admit a site through open submission followed by editorial review, or through paid submission, or by selecting resources itself against quality criteria, or by automatic means. The question of who curates is related but separate, and the possible answers are more varied than a newcomer expects: the curating may be done by paid editorial staff, by volunteer experts, by the listed parties themselves describing their own entries, by the directory’s users contributing reviews and ratings, or by software. A third dimension is the business model — whether the directory is funded, supported by advertising, paid for by the listed parties, or offered on subscription — and this matters to a reader because, as the second article noted, a directory that earns its income from the sites it lists holds an interest that does not always align with the reader’s. A last dimension is the form of the output itself: a directory may present a browsable hierarchy of described links, or a set of rich profiles carrying reviews and contact details, or something closer to a research bibliography. These dimensions are not independent of one another — a directory’s business model tends to shape its basis of inclusion, and its focus tends to shape its output — but holding them apart is what makes a fair comparison possible.
Examples by type
General web directories
The general web directory attempts to cover every subject in one structure, and the clearest examples are historical. The WWW Virtual Library, begun in 1991 and discussed at length in the previous article, is the oldest of them and, unusually, still operates as a distributed network of volunteer-maintained subject pages. Yahoo’s directory, begun in 1994, was for several years the principal way ordinary users navigated the web, and it was closed at the end of 2014. The Open Directory Project, known as DMOZ, ran from 1998 until March 2017, and at its height catalogued several million sites across roughly a million categories through the work of tens of thousands of volunteer editors; a static mirror of its final state has been kept online since under the name Curlie. These examples share a method — a browsable subject hierarchy assembled by human editors — and their history, told in the first article, shows both the appeal of that method and the reason a general directory of the whole web proved impossible to sustain against the web’s growth.
It is worth noting what these general directories looked like to a user, since the experience explains both their early appeal and their eventual decline. To use one was to start at a small set of broad headings and choose, repeatedly, the heading that seemed closest to one’s interest, descending until the listings were specific enough to be useful. For a web of modest size this was genuinely pleasant: the descent was short, the choices were obvious, and the user arrived with a sense of the surrounding territory. As the web grew, the same descent became long and the choices became fine and uncertain, and the editorial labour needed to keep the lower branches populated and current outran what any staff or volunteer body could supply. The general directory did not become a worse idea; it became an idea whose cost rose without limit, and the examples above are best read as the high-water mark of a method rather than as failures of execution. That Curlie still exists, as a volunteer-kept mirror of the Open Directory Project’s final state, is itself instructive: it shows that the appetite to maintain such a thing never quite vanished, only that the appetite was never going to be large enough, against a web of the present size, to keep a general catalogue genuinely current.
Subject and academic directories
The subject directory in its scholarly form, the subject gateway, is the matter of the fourth article in this series, and it need only be placed here within the wider field. Examples include the discipline-specific gateways funded by the United Kingdom’s eLib programme, the long-running BUBL service, and the academically oriented Internet Public Library. What distinguishes this type is not its size, which is modest, but its rigour: resources are admitted only after evaluation against explicit quality criteria, described with structured metadata, and indexed with a controlled vocabulary. A directory of this type trades breadth for trustworthiness, and it is the right instrument for a researcher or student who needs not merely relevant sources but sound ones.
Business and local directories
A large part of the directory family is organized not around subject but around organizations and, very often, around place. This type has a pedigree older than the web — the printed telephone directory’s classified section, the Yellow Pages, is its direct ancestor — and it has taken several forms online. Some online business directories catalogue companies across many sectors and many regions; the publication carrying this article, Jasmine Business Directory, is one contemporary example of that general kind. Others are local in focus and carry user-contributed reviews and ratings alongside the basic listing, of which Yelp is a widely known instance, and others again are the listing layers of larger platforms, such as the business profiles maintained within a major search engine’s mapping service. What unites the type is a focus on commerce and contactability rather than on knowledge: the question such a directory answers is which organization can be reached and relied upon for a service, and its entries accordingly emphasize contact details, location, and, increasingly, the testimony of past customers.
The business and local directory also illustrates, more sharply than any other type, how a directory’s business model bends its basis of inclusion. Because such a directory’s users are consumers looking for a service, and because the organizations being listed have a direct commercial interest in being found, the listed parties are usually willing to pay, and a directory of this kind can therefore fund itself from the entities it catalogues. That arrangement makes the type commercially durable in a way the grant-funded subject gateway never was, which is one reason business and local directories remain numerous while scholarly gateways have thinned. But it also carries the tension the second article of this series identified: a directory paid by the businesses it lists has an interest in listing them favourably, and the user has to weigh the convenience of the directory against the knowledge that its incentives are not perfectly aligned with their own. The user-contributed review, now common in this type, is in part an attempt to supply an independent signal alongside the paid listing — though reviews bring difficulties of their own, including the standing incentive to manufacture them.
Niche and vertical directories
The niche, or vertical, directory restricts itself to a single industry, profession, or field, and covers it more deeply than any general directory could. The legal directories examined in a separate article — services that list and describe law firms and individual lawyers — are a clear instance, and comparable directories exist for medicine, for the trades, for particular creative industries, and for many other defined sectors. A niche directory may organize itself by subject, by place, or by both, and its basis of inclusion ranges from open editorial review to invitation to paid listing. Its defining feature is depth within a boundary: because the field it covers is finite, it can afford the thoroughness that the first article of this series identified as the reason the niche form, alone among directory types, has remained genuinely healthy.
The niche directory’s deeper coverage also changes the kind of entry it can offer. Where a general directory, stretched across every subject, can rarely afford more than a name, a link, and a line of description, a niche directory that has confined itself to one field can record what actually matters in that field: for a directory of law firms, the areas of practice and the jurisdictions covered; for a directory of suppliers, the materials handled and the certifications held. The entry becomes less a pointer and more a structured profile, and the directory becomes useful not only for establishing that an organization exists but for comparing it sensibly against others of its kind. This is the practical payoff of accepting a boundary, and it is why a reader whose need falls squarely inside one industry is almost always better served by a good niche directory than by the relevant branch of a general one.
Table 1. Web directory types compared across the main dimensions
| Type | Scope | Focus | Basis of inclusion | Curated by |
|---|---|---|---|---|
| General web directory | Broad | Subject | Open submission and editorial review | Staff or volunteer editors |
| Subject gateway | Narrow, deep | Academic subject | Selection against quality criteria | Librarians and subject specialists |
| Business and local directory | Broad | Organizations and place | Open or paid listing; user reviews | Listed parties, users, and staff |
| Niche or vertical directory | Narrow | One industry or profession | Editorial review, invitation, or paid | Editorial staff or specialists |
What the examples have in common
Having insisted on how directories differ, it is worth pausing on what, beneath the differences, makes them all recognizably the same kind of thing — because the common core is what justifies treating them as one family at all. Three features run through every example in this article. The first is a bounded, deliberately assembled collection: a directory, of whatever type, is not the whole web but a selection from it, and the selection was made on purpose rather than stumbled upon. The second is an organizing principle: the entries are not heaped together but arranged — by subject, by place, by industry — into a structure a user can move through, and that structure, as the third article of this series argued, is authored and carries a directory’s judgement about its domain. The third is a basis of inclusion: there is always some answer, strict or lax, declared or concealed, to the question of why a given entry is present and another absent.
These three features are also, conveniently, what separates a directory from the two things it is most often confused with. A search index has no deliberately assembled collection and no real basis of inclusion beyond mechanical reach; a bare list of links may have a collection of sorts but lacks the organizing principle and the considered basis of inclusion that make a directory navigable and trustworthy. The differences catalogued earlier in this article are, in this light, variations on a shared structure rather than departures from it: every directory assembles, organizes, and admits, and the types differ only in how widely they assemble, on what principle they organize, and how strictly they admit. Holding the common core in view is what keeps the comparison coherent — it is the reason a scholarly gateway and a local business directory, for all the distance between them, can be discussed in a single article without strain.
How to read the differences
The point of laying out the dimensions and the types is practical: it lets a reader move from a need to a fitting kind of directory rather than to whichever directory happens to be most prominent. A researcher entering an unfamiliar field, who needs sound sources and an overview of how the field is organized, is served by a subject gateway, and poorly served by a paid general directory. A resident looking for a local tradesperson needs a local or business directory, where contact details and customer testimony are the substance, and would find a scholarly subject gateway simply irrelevant. The owner of a business hoping to be discovered within an industry is best placed in a well-run niche directory for that industry, where the audience is the right one. The dimensions are the questions a reader should ask of any directory before relying on it: what is its scope, what is it focused on, on what basis are sites included, who does the curating, and how is it funded.
Those same questions also let a reader recognize the type of directory worth avoiding. A directory whose basis of inclusion is purely payment, whose curation is effectively absent, and whose business model depends entirely on the fees of the listed parties is the “link directory” the first article of this series described — a directory in form but not in function, built for a ranking algorithm rather than for a reader. Nothing in the visible presentation of such a directory necessarily gives it away; what gives it away is its position on the dimensions. A reader who has the dimensions in mind can tell the difference between a directory that has something to offer and one that merely has listings to sell, and that discrimination, rather than any ranking, is what this article has tried to make possible.
A method for an unfamiliar directory
The dimensions can be turned into a short practical routine for the common situation of meeting a directory one has never used and wanting, quickly, to know what it is worth. Four checks do most of the work. The first is the inclusion policy: a directory worth trusting will usually state, somewhere, how a site comes to be listed, and a directory that says nothing at all about its basis of inclusion, or that makes payment the only visible route in, has told the reader something by its silence. The second is the entries themselves: are sites merely listed, as bare names and links, or are they described — and if described, with the particularity that suggests a person actually looked, or with the generic phrasing that suggests no one did? The third is currency: a directory whose entries point to dead pages, or whose most recent additions are years old, is not being maintained, and an unmaintained directory, as the fourth article of this series noted of subject gateways, decays into something actively misleading. The fourth is the identity of the maintainer: a directory that makes clear who stands behind it, and on what footing, offers an accountability that an anonymous one withholds. None of these checks requires special knowledge, and together they will usually place a directory on the map this article has drawn well enough for a reader to decide whether to rely on it.
Concluding remarks
Web directories cannot be sensibly arranged on a single scale of quality, because they differ in kind rather than in degree, and a ranking of them would measure a preference for one purpose over another while pretending to measure merit. What can be done, and what this article has done, is to identify the dimensions along which directories genuinely differ — scope, focus, basis of inclusion, who curates, business model, and form of output — and to place the main types against them. The general directory, the subject gateway, the business and local directory, and the niche vertical directory are not competitors for one title; they are different instruments, each suited to a different need. A reader equipped with the dimensions can do the only assessment that is actually fair to a directory: not asking whether it is the best, but asking whether it is the right one for the task at hand, and whether it was built for a reader or merely for a sale.
Future developments
The directory types described here are likely to grow less distinct in their mechanics and no less distinct in their purposes. Automatic methods will increasingly assist the population and maintenance of every type, so that the visible difference between a curated directory and an uncurated one may come to depend less on the labour involved and more on whether a stated standard is genuinely applied. The dimensions this article has set out are likely to remain the right way to tell directories apart precisely because they describe purpose and method rather than technology, and purpose and method are the slow-changing part. The most useful habit a reader can carry forward is therefore not a list of recommended directories, which will date, but the set of questions: what a directory covers, what it is focused on, how a site comes to be in it, who vouches for what it contains, and who pays. A reader who asks those questions will be able to judge a directory that does not yet exist, which is more than any ranking could offer.
References
Akerlof, G. A. (1970). The market for “lemons”: Quality uncertainty and the market mechanism. The Quarterly Journal of Economics, 84(3), 488–500.
Bakos, J. Y. (1997). Reducing buyer search costs: Implications for electronic marketplaces. Management Science, 43(12), 1676–1692.
Brin, S., & Page, L. (1998). The anatomy of a large-scale hypertextual web search engine. Computer Networks and ISDN Systems, 30(1–7), 107–117.
Espeland, W. N., & Sauder, M. (2007). Rankings and reactivity: How public measures recreate social worlds. American Journal of Sociology, 113(1), 1–40.
Hagiu, A., & Wright, J. (2015). Multi-sided platforms. International Journal of Industrial Organization, 43, 162–174.
Rochet, J.-C., & Tirole, J. (2003). Platform competition in two-sided markets. Journal of the European Economic Association, 1(4), 990–1029.
Stigler, G. J. (1961). The economics of information. Journal of Political Economy, 69(3), 213–225.
