Every industry has arguments it repeats in public. Rail and urban mobility have four that never quite end. Should the money go to new lines or to the network that already exists? Does opening regional trains to competition work? Are low-emission zones a health measure or a tax on the people who can least afford a newer car? And is the driverless vehicle a decade away, or already here? Each of these is usually argued with one document on one side and an opinion on the other. This piece does something narrower and, we think, more useful. It puts two documents side by side for each argument, one from each camp, and asks what each one actually proves. Since 25 December 2023, one of them has been settled in law, and none of them in evidence.
The first three pieces of this chapter followed the money, the slow layers a railway cannot run without, and the building sites of the year. This fourth piece reads the disagreements, because a candidate who walks into an interview at a rail operator, a transport authority or a city hall will be asked which side of them they are on.
The four great arguments in rail and urban mobility are not won by either side, because each side is measuring something different: the planner measures the condition of the network and the promoter measures capacity, the regulator counts contracts and not outcomes, the air is measured in one city and the cost of access in another, and driverless operation is routine below ground while it is still on trial above it; the professional who can name which measure each side is using is the one an employer needs in the room.
This is the fourth piece of the rail and urban mobility chapter of the CareerOn Industry Atlas. It reads eight documents, each opened and its quoted passage checked before a word was written: the synthesis report of France's infrastructure planning council, the environmental permit file of the South West high-speed project, the signed text of the EU regulation that opened regional rail to competition, the French transport regulator's summary of the rail market, a French government mission on low-emission zones across Europe, the French Senate's report on the same zones, the global metro statistics of the international public transport association, and a report to the French transport ministry on automated road vehicles. One document we wanted, the national audit office's follow-up report on regional trains and competition, did not open in any of our checks. It is not cited, and the place where it would have spoken is marked.
Build new lines, or mend the old ones
The oldest argument in French rail is usually told as a choice. On one side stand the large new lines: the high-speed project from Bordeaux to Toulouse, the new lines planned elsewhere, the ambition of bringing more cities within a few hours of each other. On the other side stands the network that already exists, much of it old, some of it slowed by its own condition, and all of it carrying the daily regional and freight trains that most passengers actually use.
The national infrastructure planning council, the body that advises the government on transport investment, does not tell this story as a choice. Its synthesis report of February 2023 sets out an order. It writes that the condition of the existing transport infrastructure calls for an effort of maintenance, renewal and modernisation that it calls the first of priorities, and it adds that its scenario does not accelerate the large projects. That is a clear position, but read carefully it is a position about sequence. The council does not cancel new lines. It puts the repair of what exists ahead of them in the queue for money.
The promoter answers in its own document. SNCF Réseau, the infrastructure manager, writes in the June 2025 edition of the South West project's permit file that the new lines will allow more frequent trains on international, national, interregional and regional services, that speed and frequency together will move travellers onto rail, and, in the sentence that matters most to this argument, that the capacity released on the existing lines will benefit regional trains and freight. In other words, the promoter argues that a new line is not the opposite of mending the old one. It is one of the ways the old one gets room back.
Set side by side, the two documents do not contradict each other as much as the public argument suggests. The council measures the condition of an asset and asks what must be spent first to keep it safe and reliable. The promoter measures capacity and asks how many trains a corridor can carry once the fastest services have their own track. Both can be right at once, and the real disagreement is about timing and about who pays while the new line is being built.
What neither document does is measure the outcome. The council does not count the capacity a new line would free. The promoter states a purpose in a permit file, which is not a measurement of freed paths or of travellers who actually changed mode. We print no figure in this section for that reason: on this argument, the honest exhibit is the two sentences, not a number. The council published a newer report this year; we have not yet read it closely enough to cite it, and it is not used here.
Competition is the law; the network is still catching up
The second argument is about who runs the regional trains. For decades, French regions bought their regional service from the national operator under direct contracts. The European Union changed that. The 2016 amendment to its public service regulation set an end date for direct awards in rail. Its text is plain: the provisions that allowed them "shall cease to apply from 25 December 2023", and any contract awarded directly in the final years before that date may not run for more than 10 years.
That is the legal side of the argument, and it is settled. The open question is what happened next, and here the documents become thinner than the debate. The French transport regulator, in its December 2025 summary of the rail market, counts the awards. By the end of 2025, it writes, 12 regional contract lots out of more than 50 planned had been awarded after a competitive process, and 4 of them went to challengers, Transdev and RATP.
Read the two together and the argument narrows. Competition is now the rule, but most of the network has not yet been through it. The regulator's figure is a count of contracts signed, not a judgement on how those contracts perform. It does not say whether tendered lines run more trains, arrive on time more often or cost a region less. That reading needs an independent audit of regional service, and the one we expected to use, the national audit office's 2024 follow-up report on regional trains and competition, could not be opened. Until it can be, any claim that tendering lowered cost or improved service is a claim this piece will not make.
The regulator also reports lower average prices on the open-access Paris to Lyon line. We leave that out on purpose: open access on a high-speed line is a different market from a regional contract, and borrowing one as evidence for the other is what this Atlas refuses.
Tendering is the law; the network is still catching up
| What the document prints | Measure | Publisher, date |
|---|---|---|
| Direct rail awards cease from 25 December 2023; late direct contracts capped at 10 years | Legal obligation and its dates | European Parliament and Council, December 2016 |
| 12 regional lots of more than 50 planned tendered by end 2025, 4 won by challengers | Regulator's count of awards | Autorité de régulation des transports, December 2025 |
Sources 4 European Parliament and Council · 1 Autorité de régulation des transports
Clean air, and who can still drive in
The third argument is the most political, and it is the one where the two sides most clearly measure different things. Low-emission zones restrict older, more polluting vehicles from city centres. Their defenders argue on air quality. Their critics argue on access: who can afford the newer car the rule requires, and what else is there for those who cannot.
The air side is best documented abroad. A French government mission, led by a former environment minister and published by the environment inspectorate in October 2023, studied how European cities made such zones acceptable. It reports London's own evaluation of its ultra-low emission zone: concentrations of nitrogen dioxide fell by 23 % between 2019 and 2022, and fine particles by 7 %. These are London's figures, as the mission reports them. They are not a measured result in a French city, and we do not print them as one.
The access side is documented in France. The Senate's report of June 2023 on low-emission zones gathers what local officials and residents told it, and the list is sharp: insufficient support from the state, clean vehicles that are financially out of reach where they are available at all, and alternatives to the car that are too thin. It then gives a single example that does more work than any slogan. In Marseille's 3rd district, which it calls the poorest in France, 52 % of vehicles are classed in the three oldest emission categories and face future bans.
The two measures never meet. One is a change in the air of one city over three years. The other is the share of vehicles in one district that a rule would exclude. Netting one against the other, as if cleaner air in London cancelled a lost car in Marseille, or the reverse, would be arithmetic without a common unit. The Senate also received a large public consultation, but replies to a consultation are testimony, not a sample, and we print no opinion figure from it. What the documents do support is a narrower and more practical conclusion: whether a zone succeeds depends on the air it cleans and on the alternatives a city builds for the people it excludes, and the second is where the jobs are.
Low-emission zones are judged on two different measures
| What the document prints | Measure | Publisher, date |
|---|---|---|
| Nitrogen dioxide down 23 % and fine particles down 7 %, 2019 to 2022 | Air quality, London's evaluation as reported | IGEDD mission, October 2023 |
| 52 % of vehicles in Marseille's 3rd district classed Crit'air 5, 4 or 3 and facing future bans | Access: vehicles affected in the poorest district | French Senate, June 2023 |
| Clean vehicles financially out of reach, alternatives to the car too thin | Documented objection from field reports | French Senate, June 2023 |
Sources 3 Inspection générale de l’environnement et du développement durable (IGEDD), mission confiée à Barbara Pompili · 2 Sénat, République française
Driverless below ground, on trial above it
The fourth argument is usually staged as a forecast: will vehicles drive themselves, and when? The documents suggest the question is already half answered, depending on where you look.
Below ground, it is answered. The international public transport association's global metro statistics, published in May 2025, count cities with at least one fully automated metro line, the highest grade of automation, in which no driver is on board. In 2023, it writes, that number "continued to increase, reaching a total of sixty". In 2000, only 13 cities had such a line. Automated lines covered 2,279 km worldwide, 11 % of all metro length. A driverless metro is not an experiment. It is a mature way of running a railway in a closed tunnel.
Above ground, the picture is different. A report remitted to the French transport ministry in April 2025, prepared under Anne-Marie Idrac, reviews the national strategy for automated and connected road vehicles. It finds that the experimental framework in force since 2018 is no longer fit for the trials it is meant to govern, because the technology has matured, and it describes the present as a transition from trials to large-scale deployment, a deployment it says is already under way in the United States and China. The report states that it does not commit the government, and its market forecasts are forecasts; we print neither as policy or as measure.
So this argument holds, but in a precise form. Autonomy is routine where the track is closed and the environment controlled, and it is still governed by trial rules where the road is open and shared. That difference is not a matter of opinion. It is a difference of environment, and it explains why the skills of safety assurance, certification and systems integration travel from the metro to the road, and not the other way round.
Driverless underground, still on trial above it
| What the document prints | Measure | Publisher, date |
|---|---|---|
| Sixty cities with a fully automated metro line in 2023, against 13 in 2000 | Worldwide count of cities | UITP, May 2025 |
| 2,279 km of automated lines, 11 % of world metro length | Worldwide length, end 2023 | UITP, May 2025 |
| French road trial framework, in force since 2018, judged no longer fit | Report to the ministry, not government policy | Transport ministry (DGITM), April 2025 |
Sources 7 UITP · 8 Ministère chargé des transports (DGITM), rapport remis par Anne-Marie Idrac
What the documents do not say
An argument piece is only as honest as its gaps, so here they are in one place. No document we read measures the capacity a new line actually frees, or the travellers who actually changed mode. No document we could open judges whether competitive tendering has lowered the cost or raised the quality of regional trains; the audit that might have done so is held. The air figures are London's, not a French city's, and the Senate's consultation is testimony, not a poll. The automation figures count lines and kilometres; they do not show that a driverless system is cheaper or safer, and the road report's market forecasts are not measures. Where a reader might expect a number in these places, we have chosen to print the gap.
There is also a counter-case to our own thesis, and it deserves a hearing. One could argue that saying both sides measure different things is a way of avoiding a verdict. The answer is that we do give verdicts, argument by argument: the first is reframed as a question of order, the second and third are narrowed to what the evidence supports, and the fourth holds. What we refuse is the false verdict that comes from adding two numbers that do not share a unit.
What this means for your career
The CareerOn view is simple. In each of these arguments, the person who adds most value is not the loudest advocate for one side. It is the person who can say, in a meeting, what each document measures and what it leaves out, and then build the next measurement that would settle the question.
For candidates, that means learning to read evidence from both camps. If you are drawn to cities and policy, low-emission zones and travel behaviour are where the analysis is thinnest and the demand is real: someone has to measure what alternatives people actually have, and how a rule changes the way they travel. The CareerOn simulation Sustainable mobility analyst: an employer travel plan rehearses exactly that work, from a travel survey to a recommendation that holds up when it is challenged.
If you are drawn to technology, the automation argument points to a specific skill. The move from closed tracks to open roads is a problem of embedded software, sensor fusion and functional safety under rules still being rewritten. The CareerOn simulation Embedded systems engineer: autonomous vehicle rehearses that discipline in a setting where the safety case matters as much as the code.
For employers, the lesson is about hiring. An operator preparing to bid for a regional contract, a transport authority defending a low-emission zone, or a supplier moving from metro automation to road vehicles all need people who can separate a legal obligation from an outcome, and a forecast from a measure. Ask candidates which number in this piece they would not trust, and why. The best answers will tell you more than a list of tools.
For leaders and investors, the arguments are a map of where evidence is missing. Each gap named above is a measurement no one has yet published with care, and the organisation that publishes it first will shape the next round of the debate.
How we read these documents
Every figure in this piece appears in a passage of one of the eight documents, and each passage was checked against the document itself before it was quoted. Each argument is shown with two documents, one from each side, and nothing is scored, netted or averaged across them. Legal obligations, counts, evaluations reported by third parties, recommendations and forecasts are labelled as such. One document, the national audit office's 2024 report on regional trains, could not be opened and is held, not cited. The full list of documents, with their dates and what each cannot prove, is in the sources below.