Privacy policy
This policy covers the Hubbub Chrome extension ("the extension"), the Hubbub API at
api.hubbub.page ("our server") and this website. In short: you don't need an account,
we don't track you, and our server never learns which threads you read.
What the extension does
On Hacker News thread pages (news.ycombinator.com/item…), the extension groups the thread's
top-level comments into topics and lets you browse them as a tree. The topic names and short descriptions
are written ahead of time by our server. Every comment the extension shows is taken from the Hacker News
page you already have open. The extension reads that page inside your browser and does not send its
contents anywhere.
The extension requests only the storage permission, plus access to
news.ycombinator.com (to show the tree on the page) and api.hubbub.page
(to download topic trees and send reports). It loads no remote code and contains no analytics.
Data the extension stores on your device
The extension uses Chrome's local extension storage (chrome.storage.local) for:
- Visit records, used to mark comments that are new since your last visit. For each thread you open with the extension, it stores the time of your visit and the highest comment id you saw. Records older than 30 days are deleted automatically. You can delete them all at any time with "Clear local history" on the extension's options page.
- Your settings, such as whether trees open automatically on big threads.
- A cached copy of the shared topic-tree file (described below), so it isn't re-downloaded on every page.
This data stays on your device. It is not synced to your Google account and is never sent to us. Uninstalling the extension removes it.
Data sent to our server
The extension talks to our server in only two ways.
1. Downloading topic trees
At most about every five minutes, while you're using it on Hacker News, the extension downloads one shared
file, https://api.hubbub.page/trees.json, that holds the topic trees for all current threads.
Everyone downloads the same file. The request is a plain download: it has no thread id, no query string and
no identifier, and it is sent without cookies or other credentials. The extension then finds the thread
you're reading inside your browser, so we never learn which thread that is.
2. Reporting a bad topic label (only if you choose to)
If you use the report button on a topic label, the extension sends: the Hacker News item id of the thread, the topic's short key, and the reason you type (up to 200 characters). This is the only request that tells us which thread it relates to, and it is sent only when you submit a report. Please don't put personal information in the reason.
As with any internet request, your browser's IP address and standard headers (such as the user agent) reach our hosting provider when you download the file or send a report. See Hosting for how they're handled.
What our server stores
| What | Contents | Kept for |
|---|---|---|
| Reports | Item id, topic key, your reason, the time received, and a review status (open or resolved). No IP address, user agent or other identifier. | Kept as a record of label reviews. Nothing in them identifies you. |
| Topic trees | Public thread information (story id and title, comment ids, comment counts) plus the AI-written topic names and descriptions. No comment text and nothing about users. | About 30 days. |
Our own server code logs only operational details such as thread ids, counts, timings and error messages. It does not log IP addresses, user agents, request bodies or comment text.
Rate limiting
To prevent report spam, reports are limited to 10 per minute per client. To enforce this, the server computes a one-way SHA-256 hash of the connecting IP address and passes that hash to Cloudflare's rate-limiting service as a counter key. The IP address and the hash are not stored in our database or written to our logs.
Hosting
Our server and this website run on Cloudflare Workers. Cloudflare, which hosts the service, processes requests (including IP addresses and request headers) to deliver them, protect against abuse, and provide the platform's request logs and metrics, which we have enabled. Cloudflare handles that data under its own privacy policy. We use these logs only to operate and debug the service.
AI processing of public Hacker News content
To create topic trees, our server itself fetches public data about busy front-page threads from the official Hacker News API and the public Algolia Hacker News search API. It sends the story's title and link plus short excerpts of public comments (roughly the first 200 characters of the top-level comments and the first 100 characters of a few replies to each) to an AI provider, Anthropic, which returns the topic grouping and labels. Commenters' usernames are not sent.
This processing uses only public Hacker News content. No data from you or your browser is ever sent to the AI provider, and users can't trigger AI requests: a scheduled job decides which threads get a tree.
No analytics, ads, selling or sharing
Hubbub has no analytics, tracking pixels, advertising or third-party scripts, in the extension or on this website. We don't sell, rent or share user data, we don't use it for advertising or credit decisions, and we don't transfer it to anyone except as needed to run the service as described above, or where the law requires it.
Hubbub does not use Google APIs. The use of information received from Google APIs will adhere to the Chrome Web Store User Data Policy, including the Limited Use requirements.
No accounts
There are no accounts, sign-ins, install ids or user ids. The extension doesn't read your Hacker News login or cookies, and it never posts anything to Hacker News.
Children
Hubbub is not directed at children under 13 and does not knowingly collect personal information from anyone, including children.
Changes to this policy
If this policy changes, we'll post the new version on this page with a new effective date.
Contact
Questions or requests about privacy: hello@hubbub.page