How we work
Method: study first, verify afterwards
How to demonstrate, before intervening, that an intervention lengthens nobody's journey.
The rule
No intervention without a prior study. The study must demonstrate that journey times for the relations concerned do not get worse, in neither the median nor the 95th percentile, at peak or off-peak, and that flows that are separate today are not overlapped. If it does not demonstrate this, the intervention is not carried out.
What the index measures
The Travel Time Index (TTI) is the same for the prior study and for the verification afterwards, so the two can be compared.
- Median time
- Door-to-door time for a set of representative origin-destination pairs, calculated for peak and off-peak periods.
- Reliability
- The 95th percentile of the same time, and its gap from the median: the margin a person must add to arrive on time almost always.
- Threshold
- No increase: the expected value after the intervention is no worse than the starting value for any of the relations considered.
How it is studied, beforehand
- The perimeter is defined, along with a sample of origin-destination pairs, including those of surface public transport, emergency services and logistics.
- Real times of the existing situation are collected (flow data, anonymous vehicle traces, surveys), not theoretical times.
- A network model is calibrated on the existing situation: if it does not reproduce the measured times, it is not used.
- The design scenario is simulated and the expected times are derived for each relation, by time band.
- The threshold is checked: no increase. The map of conflict points between flows is built, before and after: if they increase, the intervention is rejected.
- Data, assumptions, model and results are published.
How it is verified, afterwards
Once the intervention is complete, the same indicators are measured, at several points in time, to separate the initial effect from the steady state. The measurement verifies the study: it is not an experiment and does not justify an intervention already made. If journey times turn out worse than at the start, the study was wrong and the intervention is corrected or the previous state restored. The gap between forecast and measurement is published.
Transparency rules
- Assumptions are declared; data, where possible, public and reusable.
- A study that does not demonstrate the rule is published with the same prominence as one that does.
- The method is the same for our proposals and for the policies we criticise.
Status: draft method under definition. We do not yet publish studies of individual cities; when we do, it will be with the data and procedure in full.