Disassembling Trenes

Regional Depot

This essay deconstructs Spain’s current passenger railways to expose the deceptions of AVE and nation therein. “Disassembling Trenes” is the second essay in the sequence “Café Para Todos“, an exploration of the contemporary relationship between the railways and the people of Spain. The first essay, “Saving Ferroviarias“, reviews the broad policy context of Spain’s passenger railways, highlighting the residual tension between pre and post-democratic eras, the financial impetus to make the high speed network more viable, and the evolving policy paradigm of rationalisation. “Deconstructing Estaciones” provides a demographic analysis of Spain’s railway stations, that explores the unserved areas and probes the differences between regions. The fourth, “Understanding Obligación“, builds a model of the human connectivity offered by Spain’s railways, revealing the patterns between Spaniards and the democractic tension therein, with income analysis that explores the import of “Obligación de Servicio Público”. “Reanimating Regional” outlines the regional biases of Spanish railway connectivity, reassesses the role of Castilla in the national railway, and ponders the balance between actuality and perception inherent in Adolfo Suárez’s doctrine of “café para todos”.

Modelling Trains

Both infrastructure manager ADIF and the Spanish government maintain slightly different lists of railway stations and terminals, but both lists contain several hundred points which currently have no regular scheduled passenger railway service, so such infrastructure catalogues cannot be used to infer transport utility. Few Spanish public transport operators provide open electronic datasets, and most such data is limited to the larger urban areas that are not the focus of this analysis. Only recently, with the growth of channels such as Google Maps, has it become important to advertise public transport services beyond their immediate geographic locality, and many smaller operators and Autonomous Community governments still seem to lack the technical expertise or organisational will to produce interchangeable electronic data. Renfe may talk of big data and startup accelerators, but in practice can’t even manage to exchange basic schedule data within its own organisation: Regional/long distance, Iberian-gauge suburban, and narrow (metric) gauge services are each dependant on separate customer-facing internet interfaces, each of which tends to deny the existence of the services of the others. For example, passenger rail services between Gijón and Oviedo are split across all three systems, as if “Renfe” were three separate operators, which of course locally they are.

The analysis of regional railway services thus defaults to pre-electronic research methods: The manual interrogation of public timetables. Given the magnitude of the task, this was done for just one day (Friday 20 July 2018) and in one direction (away from Madrid, or broadly equivalent direction of travel). The period immediately pre-dates the signing of Renfe’s pre-liberalisation OSP contract, representing a brief period of network stability in which no major new LAV routes entered operation, and no liberalisation-related services commenced. The “day” starts with the first departures of the morning, typically at about 05:00, and continues until “end of service”, typically early Saturday morning – or Saturday daytime for overnight Trenhotel services that commence on Friday evening. Friday is one of the busiest days for longer distance rail travel so represents the network at its maximum extent (notably including a few routes which are not served daily with direct trains, such as Madrid-Águilas, or Madrid-Huelva via Medina), and also includes weekday suburban metro schedules with late evening extensions (typically until 02:00, but not throughout the night as is more common from Saturday into Sunday). July schedules do not include dedicated university services, notable only on the fringe of Barcelona, Cadiz and Córdoba. There are similar early-summer reductions in services for some Andalucían metros, but July schedules do not include the full August reductions to urban services common across Spain. In practice Renfe’s regional network changes little from day-to-day and month-to-month, so while the choice of one day and one direction will generate a few quirks, it can be considered broadly representative of regional services. The most notable differences between July and the winter months are on business-centric AVE routes, notably Madrid-Barcelona, where the July service only contains about two-thirds of the service offered in mid-September – the trains instead deployed to bolster leisure routes, such as an extra return journey each day between Barcelona and Málaga/Sevilla. While these seasonal changes alter the balance of operations and connectivity slightly, they do not radically change the number of trains operated or the places served. The observed service network linked all but three of (over 2500) stations known to have regular services. The exceptions are the dedicated university stations in Cadiz and Córdoba, which are only served in term-time, and A Pobra do Brollon (near Monforte de Lemos in Galicia), which has only one train a day eastbound (not west, the direction of travel from Madrid).

The observed service network contains all scheduled public passenger services operated on fixed rails within Spain – including metro, tram, and funicular lines where integrated into the surrounding public transport network. The only exceptions are trains that serve no local resident population and/or are priced at a “tourist” fare premium. Notable exclusions include Renfe’s Cercedilla-Cotos line (which while technically part of Madrid’s suburban Cercanías network, actually serves ski resorts) and FGC‘s Vall de Núria and Montserrat rack-railways (which are priced significantly higher than surrounding public transport, and likewise primarily serve tourist markets). Each individual service was recorded with the order of its station stops and the number of train journeys operated daily. Neighbouring pairs of stations advertised as an interchange were generally counted as a single shared station, with the principle exception of station pairs where Renfe specifically emphasise the difference between AV and local services (such as Madrid’s Atocha/Cercanías and Valencia’s Joaquín Sorolla/Nord). This method creates small inconsistencies in the count of stations, but does not affect analysis of access to services. In most cases timetable information was used directly, typically extracted from Renfe’s website by querying pairs of neighbouring towns to reveal any local services between. Train numbers were used to help identify trains uniquely across the network and avoid duplication. In a few cases, especially local seats offered on long-distance trains, both train numbers and schedule times for the same train differed slightly (notably in Galicia, as if passengers with certain ticket types board in a different manner), which required careful interpretation of schedules to identify shared trains. A small element of error is inevitable, especially when attempting to extract intricate service patterns from the Cercanías interface, which only returns journeys between the queried stations and does not provide a list of all the intermediate stations served: While some Cercanías service patterns adhere to their publicised metro-style route maps, many do not – for example, in the Asturias the majority of trains on the C2 (El Entrego) line appear to continue onto the C3 (San Juan de Nieva) line, while many C1 (Gijón southward) services routinely skip certain station stops. Cercanías’ metro-style presentation may have been copied from Madrid, but in places such as the Asturias, the practical implementation of the concept remains as varied as Renfe’s non-Cercanías Regional schedules. For metros where only service frequencies were published, the average headway was multiplied by its respective time period to give an approximate daily total. Where track engineering work had temporarily replaced trains with buses, an otherwise representative railway timetable was assumed. Non-rail modes were otherwise excluded.

Ultimately a full assessment of the role of rail should include other transport modes, and including local trams while excluding high frequency bus routes may seem inappropriate. The inclusion of trams reflects the inclusion of Renfe’s metric-gauge FEVE, the core of which operates like a low frequency segregated tramway or metro. A similar blur between subterranean metro and on-street tram is found elsewhere, with many cities presenting tram lines as “metro” services – the historic distinction between “light” and “heavy” rail no longer clear. The inclusion of metro services in turn reflects the inclusion of Renfe’s suburban Cercanías, whose operations range greatly in frequency, and on infrequent routes are directly comparable to local “regional” services: For example, “Cercanías” in Madrid includes urban services with over a hundred trains per day in each direction, while “Cercanías” in Valencia is used to describe the four trains per day (plus one shared with a longer distance regional train) from relatively rural Caudiel to Sagunt. Likewise parts of Valencia’s metro defy the popular frequency expectation of the term “metro”: For example, the southern section of Line 1 to Villamueva de Castellon carries just over 20 trains per day in each direction, a fraction of the frequency of most metro lines in Barcelona or Madrid (although both those systems contain their own frequency quirks).

The vast majority of services can be analysed as recorded. All trains are assumed to return, with observed services duplicated in reverse to complete the full service – an imperfect assumption, but one that is almost always an accurate reflection of Renfe’s service patterns. Obvious differences, notably tram routes which serve different stations in each direction, or “terminate” in a loop, have been recorded separately by direction. Services which operate in a continuous loop (Madrid metro Line 6 and Parla tram) use the start/end station shown in their respective timetables – and analysed with care. Where trains split into two portions mid-journey (which occurs on only eight long-distance commercial Renfe routes), each potion has been recorded for its full journey (as if operated separately throughout) and allocated a code to denote its unique section (to avoid double-counting over common sections in subsequent analysis). Seats offered only for local journeys on otherwise long-distance trains have been recorded in a similar manner – the local segment treated as a separate train, but denoted shared to avoid double-counting the same train in subsequent analysis. The observed service network does not contain information about connections between trains, since it does not record precise schedules. In practice low frequency regional services are not suitable for complex multi-stage journeys, while Renfe’s (non-Cercanías) products tend to emphasise direct links (not interchange), with many regional service patterns offering a wide range of different origin-destination pairs throughout the day. FEVE‘s service patterns include a few notably exceptions (El Berrón, Collanzo), but generally the absence of interchange only skew realistic journey opportunities within urban areas – which are not the focus of this analysis.

This observed network forms the basis of all the analysis contained in this sequence of essays. The next essay, “Deconstructing Estaciones“, specifically analyses stations, with the final pair of essays exploring the network’s connectivity. But before adding such complexities, a simple analysis of trains will expose the inaccuracy of many common perceptions about Spain’s railways.

Continue reading “Disassembling Trenes”


Networks of Trust in Personal Information Management

In an earlier article, I mused on the role of “thought leaders” in indirectly influencing the popularity of websites. These are further rough thoughts on the topic. Caveat: this text is not well researched.

My basic premise is this: ‘Web authors and ‘bloggers are creating trust-based filters for information. Many online writers are looking to evoke discussion and change. But most readers, most of the time, are just trying to get through the day, and aren’t too interested in discussion and change. For them, the author “sounds like they know what they’re talking about”. That creates a sense of trust, and validates the author as a reliable filter for information on that topic. Even the most objective and discerning people don’t have time to review everything themselves. They merely spend more time determining which source to trust. Likely, others will trust them, which makes the source they trust a very important actor indeed.

So we create a network of trust for information. If I want to know something about topic x I might follow the recommendation of author y, because I trust their depth of reading on that topic.

Of course, that doesn’t mean that all webmasters and bloggers are automatically trusted. Far from it. The internet or “blogosphere” is so easy to publish to, it fills up with low-grade content faster than any other media in history. Authors have to earn trust, at least from their early readers. Subsequent readers may be more prepared to trust because others are already trusting (a herd or celebrity mentality).

Why is this happening? Take Herbert Simon‘s statement that, “the rapid growth of information causes scarcity of attention.” The sentiment is repeated in Davenport and Beck’s “The Attention Economy“. We simply can’t manage all the available information any more.

Is that really a new problem? It probably hasn’t been possible to know everything there is to know since the early Victorian era. In some cases there are now technical barriers to knowledge: Simply being well educated isn’t enough to allow one to understand most cutting edge scientific developments in depth. In most cases the prime problem is volume of information: In our World of Warcraft example, more information is written than is possible for a human to read. Finding the important or useful information within can be immensely time-consuming.

Trusting people one barely knows to filter information does not automatically turn these authors into celebrities. In a few cases it may do – some readers will feel the need to trust only those who appeal to many. However, if there is a trend towards writing in narrow niches with in-depth content, rather than content with mass-appeal, an individual author may never be known to millions of people, because the topics they write about aren’t sufficiently mainstream.

Those narrow niches will similarly prevent most authors from emulating the role of pre-internet mass media, notably newspapers. They do, none the less, retain the same duty to their readers: Their readers may be inclined to trust them, but that trust will be eroded if abused. Of course, much like modern mass media, readers can still be subtly manipulated…

As I noted, Google’s biasing of sources by the number and strength links to the source. This automated approach fails to value who is creating links, so has become less valuable as the internet has become more mainstream and prone to abuse. There does not yet seem to be an effective automated equivalent of personalised networks of trust – perhaps because emulating humans is hard to do?

Implications of Google Transit in the UK

Extract from Web Trend Map. The United Kingdom’s local public transport network is likely to become part of Google Transit. Technically that should be far easier in the UK than in North America, where Google Transit was first developed: The UK has a decade’s bitter experience putting all the data together. In practice it is raising wider issues over data control and availability, that the public sector is somewhat reluctant to tackle.

This article describes how the UK’s public transport data is being integrated into Google. It questions why data is being made available based solely on the business model adopted. It explores the real value of this information, and presents a case for the liberalisation of data.

Readers unfamiliar with the topic area should read my earlier Introduction to UK Local Public Transport Data, which contains non-technical background information, and defines many of the terms used (such as “local”). The original research for this was done in June/July 2007, so may now be slightly out of date.

The illustration on the right is the Google part of a visual representation of web trends, based on the Tokyo metro map, by Information Architects Japan.

Continue reading “Implications of Google Transit in the UK”

Introduction to UK Local Public Transport Data

This article provides a basic non-technical introduction to the United Kingdom’s electronic local public transport data: The data sources primarily used to produce passenger travel information. It does not cover solely operational data, for example, financial, patronage or staff rostering.

The article is intended to provide a background for anyone wishing to understand how these data sources might be used. It was written to support my commentary on the Implications of Google Transit in the UK. The article first introduces the local public transport sector (primarily bus and rail), then explores the development of different data formats, before summarising data availability.

Continue reading “Introduction to UK Local Public Transport Data”