Privacy

Shippeth sets no tracking cookies, runs no analytics that identify you, and has no accounts. Two things are stored in your browser, both in local storage and neither of them leaving your device: your light or dark preference under shippeth-theme, and whether the stream repeats the entries that already lead the page, under shippeth-stream-repeat.

Reach measurement

We count page views on the server instead of loading an analytics script in your browser. There is no Google Analytics here and no third party involved. Two reasons: a script that a third party serves can see more than we need it to, and a large share of this audience blocks such scripts anyway, which makes the numbers both invasive and wrong.

What is stored is a counter per page and hour: the date, the hour in UTC, the path, the filters that were set, the words typed into the search box, cut to sixty characters and put in lower case, and how often that combination was requested, alongside the same counter per referring host, the host only and never the full address. The search terms are kept because they say what people look for and fail to find, which decides what we take in next. No IP address, no identifier, no cookie, and in this counter no user agent. The counter behind the outbound links stays coarser still, at one figure per day, because that one is tied to a particular item rather than to a page.

Beside it stands one more counter, of who knocked: the date, the hour in UTC, the user agent the request sent, cut to two hundred characters, a word of ours for what the request asked for, and a name out of our own list, together with how often that combination arrived. The word says whether the address was a page we serve, the feed, an address we do not serve, or one nobody reaches by reading the site, such as /.env or /wallet.dat. The name puts a browser under “reader”, a search engine under its own name, a request for such an address under “scanner”, a string that shows no browser engine under “unknown”, and a machine we do not recognise under “other-bot”. It exists because the figures on this page count readers and hold machines back, and the question of how many machines there were had no answer anywhere except in a server log.

The user agent sits beside the name because the name is a judgement and judgements turn out wrong. On 20 September 2026, 335 of that day’s 345 requests counted as readers were two machines working through a list of file names in the hope of finding a key lying about, behind a user agent that said Safari. We could see that the name was wrong and could not correct it, because the string it had been read from was gone. Keeping the string lets a corrected rule run back over the days already counted. A daily pass empties it after thirty days, once a wrong name is no longer worth correcting, and leaves the count and the name.

A user agent names a browser, a version and an operating system, a description most visitors share with many thousands of others, and it is still the most telling thing the counters keep. Beside it this counter keeps no IP address, no identifier, no cookie, no session and no time finer than the hour. Within one hour the user agent and the pages counted in that hour stand side by side, so in a quiet hour with a single reader the browser string and the pages that reader asked for, search terms included, belong together. Nothing in the counters names the person behind them, and after thirty days the string is gone and only the name and the count remain.

The page counter itself holds no trace of who did the counting: it has no session, therefore no sequence of pages, therefore no visit to reconstruct, and in it two people reading the same page in the same hour look exactly like one person reading it twice. Unknown query parameters are discarded rather than stored.

The same applies to the counter behind the outbound links: how often an item was opened or shared, per day, and nothing else.

What we count stands in full on this page and on /method. Our own reach figures are not published. They say how the site is doing, which is our business, and they say nothing about you.

What the server sees

Standard web server logs: IP address, time, requested page, user agent. They are used to keep the service running and are deleted after 14 days. They are not connected to the counters above.

Fonts

Typefaces are loaded from Google Fonts, which means Google receives your IP address when the page loads. We intend to self-host them.

Outbound links

Links to sources run through /go/<id> on this site, which records that the item was opened and then forwards you to the source. The destination stands in the link’s tooltip before you click, and the feed links to it directly.

What is stored is a row with five fields: which item, whether it was an open or a share, which channel, when, and which of the signals below showed that the click came from here. No IP address, no user agent, no identifier, no cookie. The rows cannot be linked to you or to each other, which is why this needs no consent banner and no retention period.

An open counts when it comes from one of our pages, or through a link somebody shared from here, which carries a mark of its own and a channel of its own. To tell, /go reads the Sec-Fetch-Site header your browser sets on the click, which says whether the click started on one of our pages, and stores only which of the two signals it saw. Not counted: search engine crawlers, browser prefetches, so hovering a link does not register as a visit, and a request that shows neither origin, which is how a machine that collected our links arrives.

We read the numbers ourselves, to see which items people find worth opening. They stand nowhere on the site and they influence the ranking nowhere. See the method.

Your rights

The controller is Stereum Services FlexCo, Leitzersbrunnerfeld 24, A-2003 Stockerau, Austria, reachable at office@stereum.com. The counters described above hold no IP address, no identifier and no cookie; the user agent in the agent counter describes a browser for thirty days, names nobody, and we join it to nothing. The server logs do hold an IP address, and they rest on Article 6(1)(f) of the GDPR, our interest in keeping the service running and in finding what breaks it. They are deleted after 14 days.

You may ask what is held about you, ask for it to be corrected or deleted, ask for the processing to be restricted, and object to it, by writing to the address above. A complaint goes to the Österreichische Datenschutzbehörde, Barichgasse 40-42, 1030 Vienna.

Contact

Stereum Services FlexCo, Leitzersbrunnerfeld 24, A-2003 Stockerau, Austria, office@stereum.com. See the imprint.