If you want to find small independent things and the usual surfaces are not producing any, the problem is rarely that you are searching badly. It is that you are using a route built for a different job. This is a list of routes that are built for this one.
It is organised by type of route rather than by preference, and there is no ordering inside each section. Ranking these would contradict everything else on this site.
What this is, and who it is for
A working list for anybody trying to find makers, small suppliers, independent software, personal sites or local trades, and getting the same aggregated results every time.
It is deliberately generous. Most of what is on this list is not us, several entries do things this site does not do, and if one of them serves you better then the list has done its job.
How this list is maintained, and when it was last checked
The rules, so you know what you are reading.
Every route named here was opened and confirmed working on 6 September 2026. That is the check date for this page, and it is the thing that tells you how much of the page to trust. A reading list without a check date is a claim about a web that existed whenever the author last cared.
The check date convention is the same one used across this site. A date means a named person opened the thing on that day and it was there and doing what the description says. It does not mean it is still there today. The further you are from the date, the more the entry is a historical note.
A list of live third-party services ages from the day it is written. Rings close, indexes stop crawling, hosts change what they offer and maintainers move on, and none of that is announced to us. So the commitment we make is a full re-check every three months, with the check date above replaced each time, plus an immediate correction whenever a reader tells us something here has moved or stopped. If the check date above is more than a quarter old, treat the list as stale and tell us, at corrections.
The inclusion basis is looser than the register's, deliberately. The register has a published bar with automatic exclusions and requires a named reviewer per entry. This list only requires that the route exists, that a person can use it, and that it plausibly surfaces independent work. That is a much lower bar, and it is set low on purpose because a reading list that only contained things meeting a strict register bar would be nearly empty and much less useful.
No affiliate or referral links appear here, ever. Nothing on this list paid to be on it, and nothing can.
Only routes we could open and read ourselves are named. Several sections below describe a type of route without naming an example, which means we have no example we opened and checked on the day, not that none exists. Each named route is described from what its operator publishes about it. We publish no figures about anyone's size, collection or audience.
Curated collections and registers
Collections assembled by a person or small team who decided what goes in.
Curated random-discovery collections. A maintainer keeps a list of sites worth seeing and gives you a control that shows you one. Good when you have no criteria at all and want to be surprised by something a human vouched for. indieblog.page, a personal project by Andreas Gohr, opens a random post from a personal blog. Kagi Small Web does the related job on a schedule rather than at random, showing recent posts drawn from a maintained list of independent blogs and personal sites; it is run by a commercial search company, which is worth knowing before you use it.
Personal-site directories and indexes of individual websites. Lists of people's own sites, usually maintained by someone in that community. Strong for essays, projects and hobby work; weak for anything commercial, because that is not what they are for. Ye Olde Blogroll is a hand-curated, subject-filtered list of personal and independent blogs kept by Manuel Moreale, with its sponsor disclosed on the page. Neocities hosts free personal sites and has its own browse gallery and tags over what its members have made.
Webrings. A ring links member sites to their neighbours, so following it is a way of walking a community rather than searching it. The XXIIVV webring sets out to inspire artists and developers to build their own sites and share traffic between them, and admits members by pull request. The Hotline Webring takes sites that show an individual personality and publishes what it refuses, including advertisement-focused sites and sites promoting LLMs. The IndieWeb webring connects personal sites that carry IndieWeb building blocks, and members sign in with IndieAuth or RelMeAuth.
Subject-specific collections. A single person's list covering one field in depth — a craft, a tool, a region, a technology. These are frequently the best resource in existence for their subject and are almost impossible to find unless someone tells you they exist, which is the entire problem this cluster is about. The 512KB Club collects pages around a single measurable rule it publishes: a site is listed if it holds a reasonable amount of information and its total uncompressed resources stay under 512KB.
Awesome-lists and curated repositories on code-hosting platforms. Community-maintained lists in public repositories, mostly technical, with visible edit histories so you can see whether anyone has touched them recently. The edit history is the useful part: it is one of the few places where you can check the maintenance state of a list before trusting it.
Museum-style and archival collections. Assemblies of sites gathered for their design, their history or their oddity rather than because you might buy something. Not useful for finding a supplier, and among the best routes to remembering that the web contains things that are not products.
How to read any of these. Look for the last-updated signal before you look at the contents. A collection with no visible maintenance date is not necessarily wrong, but you have no way to tell. If a collection lists when it last removed something as well as when it last added something, that is unusual and worth trusting more.
Search tools with a different index
Tools that crawl their own slice of the web rather than re-ranking someone else's results.
Independently built search indexes. A maintainer runs their own crawler with their own inclusion priorities, so the result set contains things a general-purpose index would rank far down or not carry at all. Useful precisely because it is different, not because it is more complete. Marginalia Search runs its own index and crawler software, states that it prioritises non-commercial content and uses no AI, and offers tools for browsing as well as searching. Its older address, search.marginalia.nu, redirected to that one when we checked.
Vintage- and simplicity-oriented indexes. Search over a slice selected for being lightweight, non-commercial or old-web in character. Excellent for personal and hobby pages, and not aimed at working businesses. Wiby describes itself as building a web of pages as it was in the earlier days of the internet, favours pages made by hobbyists, academics and enthusiasts, and has a "surprise me" control for people with no query at all.
Site-restricted searching on a general engine. A general engine restricted to a community, forum or platform where practitioners talk. This is a technique rather than a tool, and it remains one of the most reliable ways to reach a recommendation written by someone with no stake in it.
How to read any of these. A different index means different coverage, including different gaps. Absence from an independent index tells you nothing about whether something exists.
Newsletters and human recommenders
A person who reads widely and periodically tells you what they found. Structurally, this is a blogroll with a delivery schedule, and what the blogroll used to do explains why the attribution matters.
Subject newsletters run by one identifiable person. The recommendation carries the recommender's reputation, so it can be calibrated over time. Watch for disclosure: a newsletter with sponsored placements is fine if the placements are labelled, and much less useful if they are not.
Round-up newsletters within a trade or craft. Frequently the only place where small suppliers in a specific field are mentioned by name.
Individual writers with a links page. Some people still maintain a standing list of what they read. When you find one whose taste overlaps yours, it is worth more than any general surface.
How to read any of these. Check whether the recommender discloses commercial relationships anywhere. If they do, believe the rest more. If there is no disclosure statement at all, you are reading an unlabelled mix.
Communities where people show their work
Places where makers post what they built and other people respond.
Forums and message boards for a specific craft or trade. The recommendations are made by practitioners to practitioners, and the archive is searchable, which means a thread from years ago is often the best answer available. Lobsters is a worked example of the type in one field: a computing-focused link-aggregation and discussion site, joined by invitation, which publishes what it considers on and off topic so you can tell before posting whether your question belongs.
Chat communities and group servers. Higher-quality recommendations, much worse archives. Ask rather than search. Some communities publish their members' pages as well as talking: tilde.club gives its members space on a shared Unix machine and keeps a directory of user pages and a recently-changed list, which is a community you can read without joining.
Launch and showcase surfaces. Sites where new work gets posted and looked at. A launch record tells you something was announced. It does not tell you whether it is still running, so check the thing itself before you rely on it.
Question-and-answer threads on general social platforms. Someone asking "who actually makes these?" and getting real answers is a durable pattern, and searching for that phrasing rather than for the product often works better than searching for the product.
Mailing lists and long-running discussion groups. Older in format, still running in several fields, and often the place where the most experienced practitioners in a trade actually answer questions. Archives are usually public and searchable even where joining is not obvious.
How to read any of these. Recency is the weak point. Community recommendations age silently, and nobody goes back to update a three-year-old thread. Treat a recommendation from a community as a name to verify rather than an answer: the operator was real when the thread was written, and checking whether they still trade is a separate step you have to do yourself.
Physical and local routes that still work
The routes with no web component at all, which are consistently under-rated by people who work on the web.
Markets, fairs and craft shows. The operator is standing there. You can ask what they make, see it, and take a card. No verification problem exists at a market stall, which is a property no online route has.
Noticeboards. Library, community centre, café, workshop, church hall, veterinary surgery. Hyper-local, self-maintaining in the sense that old cards get covered by new ones, and completely invisible online.
Trade counters and suppliers. The people who sell materials know who buys them. This is the single most reliable route to a competent small trade or workshop in a given area, and it does not appear on any platform.
Word of mouth, asked deliberately. Not waiting for a recommendation, but asking a specific person who would know. The reason this works is that the recommender has met the operator, which is a standard of evidence no directory reaches.
Physical directories still printed for a trade or area. Uneven, often out of date, and occasionally the only record that a small operator exists at all.
How to suggest an addition
Send a route rather than a business. If you know somewhere that reliably surfaces independent work and is not here, tell us at corrections and it goes into the next quarterly refresh.
Two things we will not add: anything we cannot open and check ourselves, and anything with a payment relationship attached to inclusion. Our sourcing and checking rules are at our source ledger, and the argument for why every list in this space is small is at why they're all small. The rest of this cluster is at the independent web.
In Plain English
Different routes answer different questions, so use the one built for yours.
Curated collections are good when you have no criteria and want to be shown something a person vouched for. Independent search indexes are good when you want to search but away from the mainstream result set. Newsletters and individual recommenders are good when you want taste you can calibrate over time. Communities are good for practitioner-to-practitioner recommendations, with the caveat that old threads age quietly. Physical routes — markets, noticeboards, trade counters, asking someone who would know — are still the most reliable of all, because the person recommending has actually met the operator.
Check the maintenance date before you trust any list, including this one. Everything named here was opened on 6 September 2026 and was working, that date says nothing about today, and we have committed to re-checking the whole page every three months.
This site keeps its own small register, which is the shortest thing on this list and does not need to be your first stop.
Sources
Each entry below is recorded in the source ledger under the identifier shown.
- S-001 · indieblog.page — indieblog.page, Andreas Gohr. https://indieblog.page/ Checked 2026-09-06. Supports: that it opens a random post from a personal blog and is a personal project.
- S-004 · Kagi Small Web — Kagi Small Web, Kagi. https://kagi.com/smallweb Checked 2026-09-06. Supports: that it shows recent posts from a maintained list of independent blogs and personal sites, and that a commercial search company runs it.
- S-006 · Ye Olde Blogroll — Ye Olde Blogroll, Manuel Moreale. https://blogroll.org/ Checked 2026-09-06. Supports: that it is a hand-curated, subject-filtered list of personal and independent blogs with a sponsor disclosed on the page.
- S-009 · Neocities — Neocities. https://neocities.org/ Checked 2026-09-06. Supports: that it hosts free personal sites and offers a browse gallery and tags for finding them.
- S-019 · XXIIVV webring — Webring, XXIIVV. https://webring.xxiivv.com/ Checked 2026-09-06. Supports: the ring's stated purpose and that members join by pull request.
- S-003 · Hotline Webring — Hotline Webring. https://hotlinewebring.club/ Checked 2026-09-06. Supports: its membership basis and its published refusals.
- S-008 · IndieWeb webring — IndieWeb webring. https://xn--sr8hvo.ws/ Checked 2026-09-06. Supports: that it connects personal sites carrying IndieWeb building blocks and that members sign in with IndieAuth or RelMeAuth.
- S-013 · The 512KB Club — The 512KB Club. https://512kb.club/ Checked 2026-09-06. Supports: its two published inclusion rules, including the 512KB uncompressed resource limit.
- S-007 · Marginalia Search — Marginalia Search. https://marginalia-search.com/ Checked 2026-09-06. Supports: that it runs its own index and crawler software, states it prioritises non-commercial content and uses no AI, and offers discovery tools alongside search. The older address https://search.marginalia.nu/ redirected here on the day of the check.
- S-016, S-017 · Wiby — Wiby: Search Engine for the Classic Web, and its about page. https://wiby.me/ and https://wiby.me/about/ Checked 2026-09-06. Supports: its stated aim, the kinds of pages it favours, and the "surprise me" control.
- S-005 · Lobsters — About, Lobsters. https://lobste.rs/about Checked 2026-09-06. Supports: that it is a computing-focused link-aggregation and discussion community, joined by invitation, with published on- and off-topic guidance.
- S-014 · tilde.club — tilde.club. https://tilde.club/ Checked 2026-09-06. Supports: that it offers members space on a shared Unix machine and publishes a user directory and a recently-changed list.