Using Oernoe Search as a web search engine

Oernoe Search is a web search engine. That sounds too plain until you remember how many products borrow the word “search” for a feed, a chatbot, or an ads marketplace. On search.oernoe.com you type a query and get links from an index we crawl and rank. You do not get a behavioral advertising file built from that query. You do not get AdSense units in the results chrome. You do not get sponsored rows pretending to be organic ranking.

This essay is the Oernoe Editorial Team’s practical map of Search as a live technology product: what it is for, how ranking stays document-shaped, which operators How-To documents, and where the privacy boundary sits relative to the publisher site on www.oernoe.com.

## What you are opening

The Search landing page is public. Ordinary queries work without an account. Sign-in is available when you want preferences you chose — bookmarks into a private collection, for example — not because the engine needs a dossier to answer. Account creation, when you want it, happens at account.oernoe.com with free core signup and no credit card on the form.

The product promise we will stand behind is narrow: we do not use query history to build an advertising profile, and we do not sell account or search data to advertisers for their own marketing. Search still sees ordinary request data needed to return a page and fight abuse. That is operations, not an ad graph. Outbound clicks send a referrer policy that avoids handing destination sites your query string. The Search UI does not load Google AdSense.

www.oernoe.com may show ads on selected finished educational pages. Those units do not take Search queries as input. If you mash “Oernoe is private” into one network-wide sentence, you will get the funding story wrong. Keep Search and the publisher apart.

## How pages enter the index

Before a page can rank, OernoeBot has to discover and fetch it. The crawler respects robots.txt, including Crawl-delay and Disallow rules, and reads XML sitemaps where sites publish them. Crawl rates are throttled so we are not a bandwidth tax. Canonical tags and redirect chains are used to avoid scoring the same document many times under different URLs.

OernoeBot indexes public document text and hyperlink relationships. It is not a harvester of private directories, marketing address books, or origin IPs for brokerage. After fetch, HTML is sanitized: navigation chrome, footers, tracking scripts, and styling noise are stripped so the index keeps headings, body text, and useful metadata. Tokenization and stemming let related word forms meet. The inverted index maps terms to posting lists with frequencies and positions, sharded for retrieval.

None of that requires knowing who you are. Ranking starts from documents, not from a profile of the person typing.

## How ranking decides what you see

When you submit a query, candidates are scored with an enhanced Okapi BM25 approach plus structural signals. Term frequency relative to corpus rarity matters. Document length normalization keeps long pages with incidental keyword hits from crushing short, on-topic technical pages. Terms in titles, primary headings, and leading paragraphs weigh more than buried boilerplate. Inbound links from reputable technical domains contribute trust signals. Freshness helps on topics that change quickly. Performance and mobile experience factor into quality signals as described in How Search works.

What does not drive ranking: a sold advertising dossier of your past queries, a demographic vector bought from a broker, or a sponsored bid dressed as relevance. If we ever shipped sponsored results, they would have to be labeled as such and kept out of the organic claim. Today’s product statement is simpler — no AdSense, no sponsored organic cosplay.

Anti-spam and content-quality filters exist because an open web index without them becomes a junk magnet. Exact mechanics can evolve; the user-facing rule stays that Search is trying to return useful documents, not the highest bidder.

## Operators and workflows worth learning

How-To documents a small operator set that covers most careful research:

Exact phrase in quotes restricts matches to continuous sequences — useful when a technical term is also ordinary English.

site: limits results to a domain or subdomain when you already know the publisher and need pages inside it.

A leading minus excludes documents containing a term you know pollutes results.

filetype: limits to an indexed extension such as PDF when you want papers or manuals.

intitle: requires a term in the HTML title, which often correlates with pages that are actually about the subject.

Workflow for keeping good sources: run the query, open promising results, bookmark high-quality technical or research pages into your private collection when signed in, and reopen that collection from Search without publishing your bookmarks to the open web.

Workflow for entities: people and organizations can be submitted through the entity submission console with verified public canonical sources — official site, documentation, code repository. Submissions undergo human editorial review to reduce poisoning and spam listings. That is an index of public web pages plus listings people chose to submit. It is not a back door into Health notes, Chat threads, Drive files, or account settings.

## What Search is not

Search is not Mail. Mail is upcoming status only. Search is not Maps, Music, Studio, News, or OerAlerts as live How-To products. Search is not the publisher CMS. Search is not AI chat — if you want a conversational assistant, that is ai.oernoe.com, with its own training-retention rules. Search is not a promise that every page on the web is indexed, or that deletion of your Oernoe account removes a third-party article that mentions you from the index. Privacy requests about account data and takedown requests about third-party result URLs are different jobs; the privacy-requests guide separates them.

Search is also not “ad-free Oernoe.” The network has a publisher that may show ads on selected www pages. Search itself does not show those ads and does not sell queries. Say both parts.

## Account, logs, and exit

You can search signed out. Signed in, you should be able to delete preferences you saved. If a control is missing, that is a product bug — write support@oernoe.com with browser and time, never with a password. Technical logs for security and uptime may exist as Privacy describes; do not expect an account export to read like a complete diary of every query you ever typed, because Search is not building that advertising diary.

Age rules for accounts still apply when you create one: at least thirteen; under eighteen with a parent or guardian. Free core access remains the signup default.

## Independent search without hype

We will not invent index-size theater or user-count theater here. We will not claim Search makes the entire company ad-free. We will not claim ranking is magic. We will claim a live engine on search.oernoe.com, document-shaped ranking, operator support, referrer stripping, no query ad profiles, no AdSense in the Search UI, and a hard line between Search and publisher monetization on www.

Using Oernoe Search as a web search engine means treating it like what it is: an independent index with privacy constraints you can check against public guides — not a feed, not a chatbot, not an ads marketplace wearing a query box.

## A day of ordinary use without mythology

You open search.oernoe.com. You type a precise phrase in quotes because the unquoted version drowns in popular blogs. You add a site: filter when you only want documentation from one project domain. You subtract a marketing word that keeps pulling listicles. You try filetype:pdf when you need the paper, not the press release. You open three tabs, close two, bookmark one. That loop is Search doing its job.

None of that loop requires us to know your age bracket, your purchase history, or your cross-site wandering. Relevance comes from the documents. When relevance fails, the failure mode is ordinary: weak query, thin corpus coverage, spam that slipped filters, or a page too new to be crawled. The failure mode is not “we have not profiled you enough.” If results look wrong, refine the query or write support@oernoe.com with the query shape and time — still never with passwords.

Signed-out use should feel complete for ordinary lookup. Signed-in use should feel like preferences on top, not a paywall. Free core access is the company default. Labeled premium, if it appears later for Search-adjacent power features, would have to be labeled. Privacy slogans are not a substitute for a price tag.

## How Search relates to AI, Docs, and the publisher

People sometimes want Search to summarize. That is a different product job. Oernoe AI on ai.oernoe.com is the conversational assistant with ephemeral training retention rules. You can use Search to find sources and AI to draft from material you paste carefully — without pasting secrets — but the hosts remain separate. Search ranking does not become AI training. AI prompts do not become Search ad profiles. Neither becomes AdSense input on www.

Docs is where you write the research note — or a Spreadsheet or Presentation — you want to keep and share with visible permissions. Drive is where you store the PDFs you downloaded, with filters for Documents, Spreadsheets, Presentations, and uploads. Chat is where you ask a teammate what they think of a result. Tracker is where you keep personal habits and focus logs, not an ops dashboard for Search. www is where you read How Search works and Search queries and Google ads when you need the long policy form.

Keeping those hosts straight prevents a common support confusion: “I deleted my account, why is this article still in Search?” Because the article was someone else’s public web page. Account deletion is not a global web eraser. For result URLs that are not your account data, follow the privacy-requests guide’s path for third-party results, with exact URLs and reasons.

## Integrity habits for publishers who want to be found

If you publish on the open web and want OernoeBot to treat you fairly, the boring habits help: honest robots.txt, working canonicals, real titles, text that is not cloaked, sitemaps that match live URLs, and no doorways of spun junk. We will not publish a guarantee that every good page ranks above every bad page. We will say the ranking stack is built to reward document quality, structure, citations, and freshness — and to refuse query-profiling as a substitute for those signals.

Entity submissions need canonical evidence. Editorial review exists because open submission without review becomes spam. That is quality control, not a secret loyalty program.

## What we refuse to say about Search

We refuse “zero tracking” as a network-wide promise. We refuse “ad-free Oernoe” while www may show ads on selected finished pages. We refuse claiming Search sells queries. We refuse claiming Search is Mail or Maps. We refuse millions-of-queries theater. We refuse implying that BM25-shaped scoring makes human judgment unnecessary for spam.

What remains is usable: an independent web search engine on search.oernoe.com, public query access, optional account preferences, documented operators, crawler manners, document-shaped ranking, referrer stripping, no AdSense in the Search UI, and a bright line against query advertising files. Use it as a search engine. Hold us to those sentences — and only those sentences.