A transit map that needs explaining has already failed

There is a settled way of talking about network legibility. It is worth asking how much of it survives contact with the detail.

Legibility is a use condition, not a nicety

A person who cannot work out how to make a trip will not make it, whatever the service quality is once they board. Regular commuters learn one journey and need little help, while occasional users, visitors and new residents depend entirely on the network being readable. Those occasional trips are a large share of total potential ridership and the share most easily lost to other modes.

A network that requires a phone application to be usable has excluded everybody without one and made itself invisible to everybody who has not looked. Legibility is therefore a form of capacity, since it determines how much of the potential market can actually reach the service.

Branches and variants are the main destroyer

A single route that splits into several destinations means the passenger must check which vehicle they are boarding rather than simply boarding. Variants that run only at certain times, or divert on certain days, multiply the information a passenger must hold to use the route confidently.

These arrangements usually accumulate over decades as responses to individual requests, each reasonable at the time it was granted. Consolidating them raises real objections from the small groups who use each variant, and those objections are specific while the benefit is diffuse. Network redesigns that simplify routes reliably produce that political shape, which is why they are difficult regardless of their merits.

Metro maps taught the wrong lesson to buses

Rail diagrams simplify geography because a passenger cannot leave the system between stations and does not need to know where they are. Bus networks are different, since passengers walk to stops, cross streets and need to relate the route to the actual street pattern. A schematic bus map that distorts geography can leave a passenger unable to tell whether two stops are adjacent or a long walk apart.

Good bus maps keep geographic accuracy and reduce complexity by showing fewer routes, typically only the frequent ones. The distinction is between simplifying the drawing and simplifying the network, and only the second one changes what people can use.

Frequency maps show usefulness

A conventional route map shows every service equally, so a route running a few times a day looks like one running every few minutes. A frequency map distinguishes them by line weight or colour, letting a reader see immediately where they can travel without planning.

Cities publishing frequency maps have generally found them well received, because they answer the question passengers are actually asking. The downside is that infrequent routes become visibly marginal, which creates pressure that some operators would rather avoid.

Publishing the map honestly is nonetheless the precondition for arguing about which corridors deserve investment.

Naming and numbering carry information

Route numbers that group by corridor, direction or type let a passenger infer something about an unfamiliar service before looking it up. Where numbers are historical accidents, each one must be memorised individually, which is a cost paid by every occasional user. Renumbering is disruptive for regular passengers and pays back over years, which makes it a hard sell in any single budget cycle.

In practice, consistent naming of interchanges matters just as much, since the same place called two different things on two modes breaks the connection. These are small decisions with compounding effects, and they are usually made without much deliberation.

Testing a network for legibility

Ask somebody unfamiliar with the city to plan a specific trip using only what is displayed at a stop, and watch where they hesitate. Check whether a stop shows the routes serving it, their destinations and a local map, since many show only a number.

Across a network, see whether the same corridor is described consistently across maps, timetables, vehicle displays and signage at interchanges. Count the route variants on a single number, as more than two is usually a sign the network has accumulated exceptions. The test is not whether an expert can decode the system but whether a stranger can use it once without asking anybody.

The takeaway

Hand the map to a stranger. Whatever they cannot work out is the network design problem.

Measure the walk, not the map.