Who writes and who publishes?
What Ridelta publishes is written by the team that runs the platform and signed by the brand, not by a pseudonym or an outside agency. There are two paths and both go through human review. The first is code: the guides for each province, the knowledge-centre pieces and the glossary terms are written in the project's repository and come in through a change request that someone reviews, with every automatic check green, before reaching the website. The second is the panel newsroom: whoever operates the platform writes an article, sends it for review and only a person with administration rights can publish it. A draft is never published without review, and what has been published is only withdrawn; it is not deleted.
Artificial intelligence may propose a draft, and only that. The draft generator is switched off by default; if the owner switches it on, a person asks for a draft with a topic and an audience, the text comes in as a draft marked with its origin and goes through the same filter as any other. The machine does not publish, does not correct what is published and does not sign: every piece seen on the website has been read and approved by a person who answers for it. If a piece was written with the help of an automatic draft, its revision history says so.
Where do the figures come from?
No figure that can change is written by hand into a text. The prices, distances and times of a route are calculated by the route engine when the page is rendered, from the tariff in force and the stored distance, and the text only leaves a slot for them to fill; the commission charged to partners comes from the setting published in the panel, the same one they see. That is why a guide can say what it costs to go from a village to the airport without anyone having to remember to update it when the tariff changes: the number changes at the source and it changes on every page at once.
Automatic tests enforce the rule before each publication: they search the content for a hand-written amount in euros, a waiting or courtesy time in minutes, a commission percentage or a promise with a figure, and if they find one the publication does not go out. The pieces of the global domain, which apply to any province, carry no figure for price, waiting or commission: they explain the method and refer to the calculator on each area's site, which is where the exact number for an address, a date and a time lives. The observatory's figures are a different matter: they are data from the public administration, each with its source, its link and the date it was consulted, and a region without published data has no page.
Which sources are accepted and how are they cited?
Every guide and every piece cites its sources at the foot, with the name of what was consulted, the link and the date of consultation, and that date cannot be later than the text's last revision: the test checks it. Public, verifiable sources are accepted: the official state gazette and the regional gazettes, ministries and their registers, state agencies, European legislation published in its official journal, public transport operators, airport and station managers, health services, town councils and statistics offices. A fact that cannot be linked to one of those sources is not published as a fact; at most it is told as the team's experience, and it says so.
Sources are also marked up for machines: each article's structured data carries the list of citations with their name and link, so that a search engine or an assistant can trace where each statement comes from without reading the whole page. When a rule changes, the piece that cites it is reviewed and its revision date is updated; until it is reviewed, the validity notice that closes each piece recalls that what it says is what the source said on the day of consultation. What competitors write, what circulates on social networks or what a customer tells us are not sources: they can be a starting point for finding the source, never the fact itself.
What is never published?
Nothing is promised that the code does not do. No page says how many jobs a month a driver will do, how much they will earn, that there is always a free car or that the car arrives in five minutes: automatic tests search for those phrases, in Spanish and in English, and block publication if they appear. No thin pages are published, made from a template by changing the name of the town, because search engines treat them as filler and because they help no one: every province's own guide has at least nine hundred words in Spanish, a complete English version, a direct answer at the top and dated sources, and no paragraph is repeated between two guides.
The brand's global domain publishes nothing local: not an area, not a town, not a route, not a destination. What belongs to a province is told once, on that province's site, which is its canonical address, and the test on the global pieces checks against the catalogue that none of them names anything local. Nor are invented opinions or reviews that no real customer left published, nor a machine-generated text that a person has not reviewed and approved, nor a fact whose source cannot be linked. And no content is published to deceive a search engine: nothing hidden, nothing deliberately duplicated, nothing that tells the machine one thing and the person another.
How is an error corrected?
Anyone can point out an error from the contact page, giving the address of the page and the fact they believe is wrong. The team checks the fact against its source; if the source proves the reporter right, the text is corrected, the piece's last-revision date is updated and the sitemap publishes that new date, so that search engines read it again. If the source has changed since it was consulted, it is consulted again and the source date is updated too. An error that changes the meaning of what was said is not corrected silently: the piece says so, with the date, in its validity notice.
Nothing that was published disappears without trace. Newsroom articles keep every revision with its status, its author and the reason, and what is withdrawn still exists in the history so that what was said, and when, can be explained. The observatory's figures are corrected or withdrawn from the panel, with a reason, but not deleted. The pieces in code have their history in the project's repository, change by change. If a correction affects a price or a condition of the service, what the calculator and the terms in force publish at the time of booking prevail, not the text of a guide.
What do the dates on each piece mean?
Every guide and every piece carries two real dates: the publication date, which is the day it first appeared, and the last-revision date, which is the day someone read it again and approved or changed it. Both go in the article's structured data and the second is the one the sitemap declares as the modification date. They are not touched to look recent: a piece that has not been reviewed keeps its date as time passes, because a false date is a page's first lie and the one a search engine detects first. The table below summarises what the platform publishes and when each block was last reviewed.
This policy also has a date and is also reviewed. It was written with what the code does today, and every time a rule changes — a new check, a source that is no longer accepted, a new path to publish — it will be updated here before anywhere else. The pieces link to it as their publishing principles and their corrections policy in their structured data, so that whoever reads a guide can know, with one click, how it was written and how to ask for it to be corrected.
What we publish and when it was last reviewed
| Block | Pieces | Last reviewed |
|---|---|---|
| Each province's own guides | 74 | 27 October 2026 |
| Regional guides | 12 | 15 September 2026 |
| Knowledge-centre pieces | 24 | 27 October 2026 |
| Glossary terms | 92 | 27 October 2026 |
The dates are the ones each piece declares in its structured data and in the sitemap; newsroom articles and observatory figures carry their own on their own pages.