Privacypolicy

Tamborradata is a website that offers statistics about the participants of the children companies of the Tamborrada of Donostia / San Sebastián, and a search tool to check in which editions a participant appears.

The controller of the data described in this policy is the owner of Tamborradata. For any privacy-related matter, including a request to remove a name, you can write to contact@tamborradata.com.

We process two kinds of data, very different in nature.

Participant data. This is what appears in the published lists of the Tamborrada children companies. For each participant we store:

• Their first name and surnames, as published and in a normalized version.
• The school or company they belong to.
• The year of the edition in which they appear.
• The address of the page where it was published, so that we can link the source.
• A calculated split between first name and surnames, which we use for the statistics.

We do not store age, date of birth, sex, images, addresses or any contact details of participants.

Visitor data. We do not ask you to register, there are no user accounts and no contact forms. Only the technical data described in the cookies and analytics section, and the data generated by the use of the search, are processed.

Participant data is not provided to Tamborradata by the participants or their families. We collect it from a single public source: the pages that Diario Vasco dedicates to the Tamborrada children companies.

The collection process is automatic and has two steps:

• Finding the pages. We check the section listing on Diario Vasco, a search engine (through the SerpAPI API) and the historical copies kept by the Internet Archive for the oldest editions.
• Reading the names. A program reads the school name and the list of participants from each page. There is no manual data entry.

The data is not official. It depends on what the newspaper published at the time and may contain errors, omissions or differences from the actual list of each edition.

Once the names have been read, we sort and study them in order to offer statistics and searches. These are the steps:

• Cleaning. Lines that do not look like a person name (with digits, addresses, or too short or too long) are discarded and the format is unified, for example "Surnames, First name" becomes "First name Surnames". Accents and capital letters are also removed so that names can be compared.
• Schools. Each name is linked to a school. When the same school is written in several ways, we group them by similarity.
• First name and surnames. The published text does not tell first names from surnames, so we split it with a statistical method based on how often each word appears as a first name or as a surname across the whole dataset. It does not use artificial intelligence. The split is probabilistic, and doubtful cases are left out of the statistics and the search.
• Duplicates. If the same name appears several times in the same school and year, it is stored only once.

There is no person identifier. The same person is recognised across editions only because their normalized name matches, so two different people with the same name, school and year may appear as one.

With the processed data we calculate statistics for each edition and for all of them together: most common first names and surnames, evolution over the years, newcomers and disappearances, diversity and concentration, and school rankings.

• Rankings and charts show aggregated data: how many times a first name, a surname or a combination of both appears.
• Rankings of first name and surname combinations only include pairs that appear at least twice and count participations, not distinct people. The full name of a person is never published in a statistic.
• Statistics are recalculated every time a new edition is added or a record is removed.

AI-written summaries. Some explanatory texts of the statistics are written by an OpenAI language model and translated into several languages. The model receives only aggregated data (rankings, series and values) and never the participant rows. It is explicitly told not to infer kinship, origin or other characteristics of people. Like any automatically generated text, it may contain inaccuracies.

The search lets you check in which editions a participant appears. To prevent it from being used to browse whole lists, it requires two pieces of information at once.

• What it asks for. The full name (at least three words, at most 120 characters) and the school, chosen from a list.
• What is compared. The text is converted to lowercase without accents and compared with the participants of that school. Exact and compatible matches are accepted (for example, without the second first name or the second surname), but never without the first surname.
• What is filtered. Only records whose split into first name and surnames could be resolved reliably are taken into account.
• What is shown. The name as it was published, the school and the years in which it appears, each with a link to the page where it was published. Exact matches are shown first and then the most recent ones.
• What is not done. No comparisons or scores are generated about specific people.

What happens to what you search. The name is sent to the server in the body of the request, not in the page address, and our code does not store or log it. To limit abuse, we restrict the number of searches per IP address: we use an encrypted fingerprint of the IP that is kept only in the server memory, not the plain IP. The hosting provider may log technical data about requests, as explained in the providers and third parties section.

The companies the data is taken from are children companies, so most of the people who appear are minors or were at the time of publication. We say it clearly because we believe you should know.

We only store what a media outlet had already published (name, school and year), with no additional data, and we do not try to identify, locate or contact anyone. The legal basis for this processing is the legitimate interest in offering statistical and cultural information about a public tradition, weighed against the rights of the people affected.

If you appear on Tamborradata, or you are their mother, father or guardian, and you prefer not to appear, you only need to ask. You do not have to give a reason. See how to do it in the rights section.

Tamborradata does not use advertising or tracking cookies, and does not include third-party tools to build profiles.

• Language cookie. The website may store a cookie called NEXT_LOCALE with the chosen language (Spanish, Basque or English). It is technical, needed to remember your preference, lasts for the session and contains no personal data.
• Visit analytics. We use Vercel Analytics to know, in aggregate, which pages are visited. According to the provider documentation, it does not use cookies and does not track you across different websites.
• Performance. We use Vercel Speed Insights to measure page loading speed (technical metrics such as load time, device type or browser).

We do not store your searches in the browser beyond the memory of the open tab, which is cleared when you close it.

To work, Tamborradata relies on these providers:

• Vercel. Hosts the website and provides visit analytics and performance metrics. As with any hosting, its servers may log technical data about requests, such as the IP address, the requested page or the browser.
• Supabase. Hosts the database where participants and statistics are stored.
• OpenAI. Writes and translates the summaries of the statistics. It receives only aggregated data, never participant names (see the statistics section).
• SerpAPI and Internet Archive. Used in the process of finding the source pages. They do not receive data from Tamborradata visitors.

We do not sell or hand over data to third parties for commercial purposes. Some of these providers may process data outside the European Economic Area, with the safeguards they offer in their own privacy policies.

You have the right to:

• Know whether we process data about you and access it.
• Ask us to correct inaccurate data, for example a misspelled name.
• Ask us to delete your data (right to erasure).
• Object to the processing or ask for it to be restricted.

How to remove a name. Write to contact@tamborradata.com stating the name as it appears, the school and, if you remember it, the year. If the request is made by a mother, a father or a guardian, it is enough to say so. We may ask for a minimal check to make sure the request is legitimate.

We will reply without undue delay and, at most, within one month. When a record is removed it disappears from the search and the statistics are recalculated so that its contribution no longer counts.

Removing a name from Tamborradata does not remove it from the original page of the newspaper or from the Internet Archive copies, which are outside our control. If you are not satisfied with our reply, you can lodge a complaint with the Spanish Data Protection Agency (aepd.es).

We keep participant data while the project remains active, since its purpose is precisely to compare editions over time, or until its removal is requested.

The database is only modified by our internal update process, and the website accesses it only from the server. Access keys are not exposed in the browser.

No system is infallible. If we detect an incident affecting personal data, we will act in accordance with the applicable regulations.

We may update this policy when the way we process data changes, for example if we add a new source or a new tool. The date of the last modification appears at the top of this page.

If you have any question about how Tamborradata works or about your data, write to us and we will explain it.