Strip search down to its skeleton and it is two decisions performed at planetary scale. Decision one: what exists — the crawl and the index, the library of pages a query can possibly return. Decision two: what surfaces — the ranking, the ordering of that library against your words. Everything a family experiences on a results page — what appears, what never appears, what sits first — is downstream of those two decisions. And here is the whole kosher-search problem in one sentence: for the entire history of the frum internet, both decisions were made by companies for whom this community's standards were, at best, a settings option.
What an overlay is, architecturally
Once you see the two decisions, every product in the kosher-search toolbox reveals its true shape: an overlay — an intervention performed after both decisions have already been made by someone else.
- The safety modes and their DNS locks are post-ranking veto: Google decides what exists and what surfaces; the mode then vetoes some of what surfaced. Its ceiling is its position in the pipeline — it can only subtract from decisions it never participated in, which is why the vendor's own documentation hedges its accuracy.
- The human screening rooms — Israel's NetFree, with staff reviewing images around the clock — are post-surface review: the most labor-intensive veto ever constructed, standing between families and a results page whose composition they cannot influence.
- The children's portals are pre-query confinement: rather than veto results, they shrink the library to a curated corner — honest, and unlivable as a household's whole search.
- Even the values engines mostly rent: an engine "powered by" another's index has outsourced decision one, keeping only a say in decision two.
Every overlay shares one property, and it is the property that matters: the upstream can move. The index refreshes, the ranking model updates, the results page reshuffles — on the owner's schedule, for the owner's reasons — and every veto, screen, and lock downstream must chase. The overlay era's endless maintenance is not bad engineering. It is the physics of standing downstream.
“An overlay can only subtract from decisions it never participated in. Owning the engine means making the decisions.”
kolbo.life
What owning the engine changes
Now read the kolbo.life homepage's description of KolBo Search as an architecture statement: "Not a filtered feed sitting on someone else's index — our own engine, built on open sources, returning clean, relevant results tuned for how this community actually searches."
Our own engine is decision one and decision two, repatriated. The index is built — "on open sources," the open web's raw material rather than a rented feed — by the same party that answers to this market; cleanliness becomes a property of what exists in the library, not a veto racing what surfaced from someone else's. Tuned for how this community actually searches is decision two, made fluent: ranking that treats communal queries as normal rather than noise. And the placement completes it: shipped inside "the complete operating layer for kosher devices," the engine is the default front door rather than a destination requiring enforcement — the same built-versus-bolted logic that runs through the browser story and the whole suite.
None of this makes overlays foolish in retrospect — they were the only tools downstream residents could build, and the community built the best of them. It makes them, finally, optional. The honest limits travel with the claim, as always: the homepage states no index size, no coverage details, no standalone availability — the engine ships within the licensed device suite, and where the homepage is silent this page is too. But the architecture question this essay set out to answer is settled by the sentence itself: a filtered feed is someone else's decisions, edited. An engine is the decisions, owned. The full story — the workaround stack, Koogle, the kid-safe aisle — is the pillar.
Frequently asked questions
What's the difference between a search engine and a filtered feed?
Position in the pipeline. An engine makes the two decisions that constitute search — what enters the index, and what surfaces per query. A filtered feed intervenes after both decisions were made elsewhere: vetoing, blurring, or screening someone else's results page, forever downstream.
Why do overlays require constant maintenance?
Because the upstream moves: indexes refresh and ranking models update on the owner's schedule, and every downstream lock and screen must chase. That is physics, not vendor failure — the only exit is owning the decisions.
What does "built on open sources" mean?
The homepage's phrase for the index's raw material: the open web, rather than a rented results feed. The exact composition is the builder's to state — the homepage says no more, and neither do we.
Is human image-screening an overlay too?
Yes — the most heroic one: staff reviewing what someone else's engine surfaced, in real time, indefinitely. It exists precisely because the community stood downstream of an index it didn't own; an owned engine is what retires the screening room.
- kolbo.life — founder-approved product source; all KolBo claims quoted verbatim (verified July 2, 2026)
- Google — SafeSearch settings — the post-ranking veto, documented (verified July 2, 2026)
- techkosher.org — NetFree — the human-review overlay at communal scale
- Kiddle — about — pre-query confinement, in its own words (verified July 2, 2026)
- Swisscows — the rented-index pattern in values engines (verified July 2, 2026)
Protection for the device already in your pocket
KolBo Secure protects any iPhone or Android — tamper-resistant enforcement, a self-service portal, and real human support. Starting at $14.99/month.
Secure a deviceEnrollment, configuration, and billing in one portal — minutes, not appointments.