HN Debrief

Why Book Corners won't sync contributions back to OpenStreetMap

  • Open Source
  • Maps
  • Data Governance
  • Developer Tools

The post is from the creator of Book Corners, a site that maps public book-sharing boxes like Little Free Libraries. They had considered syncing user-submitted locations back into OpenStreetMap, then backed away after digging into OSM’s import rules. The blocker is not technical integration. It is everything around it. OSM treats a third-party app that collects data and then feeds it in as an import pipeline, which brings requirements for clear licensing, a documented process, community review, quality control, rollback plans, and ongoing maintenance. The author’s conclusion is that this is a real operational commitment, not a side feature.

If you build on top of a shared open dataset, decide early whether you are a view over that source, a separate database, or a feeder of structured hints back into it. If you want contributions to flow upstream, direct user edits into the upstream system from day one or plan for a much heavier operational burden than the sync code itself suggests.

Discussion mood

Mostly sympathetic to OSM’s conservatism. People found the rules annoying but justified because bulk geodata mistakes, spam, and licensing problems are expensive to unwind, and many thought the real mistake was trying to sync a separate app database upstream after the fact.

Key insights

  1. 01

    Separate app data from canonical map data

    The clean architecture is to keep the shared map as the canonical record for the thing’s existence and location, then attach your own extra fields locally. That lets your app store what OSM does not care about while avoiding the endless drift that comes from maintaining a competing base map. The bouldering example made the trade-off concrete. The site keeps its own metadata for a niche audience, but new objects must first exist in the upstream database before they are imported and enriched.

    If your product depends on a shared map layer, design your schema around foreign references plus local annotations. Do not let your app become the place where core geographic objects are born unless you are prepared to own permanent curation work.

      Attribution:
    • jasonkester #1 #2 #3
  2. 02

    OSM is weak as a reference key

    A recurring pain point is that OSM features are not stable identifiers for outside datasets. Mappers can split, merge, or replace features during normal editing, which breaks downstream references even when the real-world object has not changed. That makes OSM awkward for niche overlays like canoe navigability or noise data that need long-lived links to geographic features plus their own domain-specific attributes.

    Do not assume an OSM object ID can serve as your durable primary key. If you need long-term joins between your data and map features, budget for your own identity layer and reconciliation logic.

      Attribution:
    • mikeocool #1
    • jasonkester #1
    • arjie #1
  3. 03

    Direct-to-OSM submission changes the compliance burden

    The legal and process burden drops a lot when your app is just an editor used by the contributor, rather than a service collecting data and later exporting it in bulk. Several comments argued that if Book Corners authenticated users with OSM and sent edits straight to OSM, those contributions would look like normal user submissions instead of an import feed. That does not remove moderation or tagging issues, but it avoids turning the app operator into the sole party responsible for a parallel ingest pipeline.

    If upstream contribution is a product goal, integrate with the upstream account system early. Moving user edits directly into the canonical platform is often far simpler than building your own staging database and trying to reconcile it later.

      Attribution:
    • WhyNotHugo #1
    • maxerickson #1
    • marisen #1
  4. 04

    Notes and tasks fit uncertain field reports

    There is already a lighter-weight channel for useful but unverified map information. OSM Notes and MapRoulette let outside apps pass along likely changes with photos, links, and descriptions, while leaving the final edit to mappers who validate on the ground or through approved imagery. That preserves signal without granting full trust to an external dataset.

    When you have credible geospatial hints but not import-grade data governance, ship them as reviewable tasks instead of edits. You will get slower uptake, but you avoid licensing and rollback risk while still helping the map improve.

      Attribution:
    • florkbork #1
    • Maxious #1
    • progbits #1
  5. 05

    Publishing the dataset unlocked community action

    The author’s mid-thread change mattered more than the original blog post. By releasing the full library database as ODbL GeoJSON, they removed the biggest blocker for anyone else who wants to validate or import the missing book boxes into OSM. That turns Book Corners from a dead-end fork into a usable feeder of open data, even without an official sync feature.

    If you cannot run the upstream contribution process yourself, at least publish exports under compatible terms with enough structure for others to act. Open release can be the difference between a silo and a contribution pipeline.

      Attribution:
    • Rygian #1
    • andreagrandi #1 #2

Against the grain

  1. 01

    High coverage can beat pristine but sparse maps

    For accessibility use cases like disabled parking, incomplete but broad coverage may be more useful than a smaller set of highly verified points. That pushes against OSM’s bias toward cautious inclusion and highlights a product truth OSM does not optimize for. Users often need probabilistic help now, not perfect data later. A commenter working on Dutch accessible parking argued that transparency about confidence can make lower-quality data still actionable.

    If your users face real-world access constraints, consider shipping confidence-scored coverage in your own product even when the upstream map will not accept it yet. Make the uncertainty visible so people can judge the risk.

      Attribution:
    • Moghammed #1
    • berkes #1
  2. 02

    Strict ingest rules also keep POI data stale

    The same guardrails that protect OSM from junk also make place data laggy and incomplete. Some argued OSM should separate storage from display more aggressively, allowing lower-trust submissions into a broader pool and letting map consumers filter by verification level. That would trade a single hard quality bar for multiple quality tiers, closer to how modern data products handle noisy inputs.

    If you run a map-adjacent product, do not assume one database has to serve every trust level. A tiered model for raw signals, reviewed facts, and user-facing display can unlock more contribution volume without collapsing your quality bar.

      Attribution:
    • greyb #1
    • londons_explore #1

In plain english

CoMaps
A mapping app mentioned in the comments as a tool that can make OpenStreetMap contributions easier.
EveryDoor
A mobile app for quickly editing shops and amenities in OpenStreetMap while in the field.
GeoJSON
A common text-based format for storing geographic data such as points, lines, and polygons.
MapComplete
A specialized editor for OpenStreetMap that presents topic-specific map editing interfaces.
MapRoulette
A tool that turns mapping problems into small review tasks for volunteers to validate and fix.
OAuth
An authentication standard that lets one app ask a user to log in with another service without sharing their password directly.
ODbL
Open Database License, the share-alike license used by OpenStreetMap data and some related open datasets.
Organic Maps
An offline map app mentioned in the comments that supports contributing data to OpenStreetMap.
OSM
OpenStreetMap, a collaborative open map of the world that volunteers edit and that anyone can use under an open license.
StreetComplete
A mobile app that asks simple questions to help people improve OpenStreetMap data while they are out walking around.

Reference links

OSM contribution and review workflows

  • OpenStreetMap Notes
    Suggested as a lightweight way for Book Corners to send likely locations back as reviewable signals instead of direct edits

Examples of single-source-of-truth niche maps

  • Betty Beta
    Example of a niche site that imports from a canonical upstream climbing database rather than creating new core records itself
  • Bleau.info
    Used as the canonical upstream database in the climbing example about avoiding data drift

Street-level imagery platforms

  • Panoramax
    Mentioned as an open street-level imagery service already handling blurring and some object detection for mapping use

Project data export

  • Book Corners post
    The original post where the author explains the decision and later notes the ODbL GeoJSON export in comments