Look up an emoji. Find a charging station. Check whether a tool exists for a specific OSINT task. Find out if there’s a giant fibreglass banana on the way to wherever I’m driving. What they’ve all picked up over the years, is ads.
Emojipedia is the clearest case. It’s been running display ads since at least 2015, and since Zedge bought it in 2021 it has only gotten heavier: banners, interstitials, and a layout that makes you hunt for the emoji under the monetisation. PlugShare, the charging map most EV drivers end up with, now sells a dedicated ad-free subscription for $1.49 a month or $14.99 a year, which tells you plainly what the free tier is for. OSINT tool directories are their own genre of bad, because “free tool aggregator” and “ad-tech dumping ground” have always overlapped. None of this is a scandal. It’s just what happens to a reference site once it needs a business model.
Building instead of tolerating
The apps I use for these four things don’t need a business model, because nobody else uses them. So I built four small, ad-free replacements, using Hyperagent and Claude Code. The split was consistent across all four: Hyperagent handled the fetching, pulling structured data out of the open sources each project depends on, and created the initial app, turning that data into a site with its own query logic rather than a wrapper around someone else’s API. Then Claude Code fixed some presentation and user experience issues, added additional features, and polished the overall look and feel.
Emojisaurus, an emoji reference with no interstitials
emojisaurus.me is a straight search tool, tracking Emoji 17.0. You can search by name, subgroup, codepoints, UTF-8 or UTF-16 encoding, or version, sort by relevance, name or subgroup, and filter by skin tone or qualification status (fully-qualified, minimally-qualified, unqualified, or all of them at once). It’s the kind of dense, query-syntax interface a reference tool should have, and the kind Emojipedia used to have before it became mostly ad inventory with emoji-shaped content between the banners.
A charging map without a subscription
map.electricvehicle.life exists because I got tired of an EV charging app selling me the absence of ads as a feature. It’s tied to my EV blog, and I’m using it to work out charging strategy for a December road trip down through Melbourne and Adelaide. It doesn’t need to sell an ad-free tier because there was never an ad tier to begin with. To be honest, open source data only takes you so far, and the map is missing a lot of chargers that are in the PlugShare database.
An OSINT tools directory that’s just a directory
osintexplorer.halans.dev is a browsable list of the OSINT tools, sorted the way I think about them rather than the way an SEO-optimised aggregator wants to rank them. Given how much of that space runs on ad networks of dubious quality, having a version that’s just a list, with nothing trying to load in the background, is most of the point.
The Big Things map, and what honesty costs
bigthings.downunder.blue is the one that went furthest. It maps Australia’s giant roadside sculptures, the fibreglass fruit and oversized fauna that towns built to advertise themselves. There’s no official register of these things, only a 2023 academic census that counted 1,075 of them nationally. The map currently shows 475, and says so on the front page rather than pretending to completeness.
Of those 475, 313 (66%) are pinned to the sculpture’s actual location; the rest sit at their town’s centre, marked with a dashed outline, because that’s all the source data gives. Twenty-six sculptures that have been demolished or removed stay on the map instead of being quietly deleted, because the ghosts are part of the record. And there’s a running log of 25+ corrections, each with its own source, catching things like a transposed longitude digit that put one sculpture 350 kilometres from where it stands. The data comes from five open sources, Wikipedia, Wikivoyage, OpenStreetMap, Wikimedia Commons, and two community catalogues, each cited per row. The whole thing, including 301 photographs and the map tiles, is served from its own domain (on Cloudflare) instead of hotlinking anyone else’s infrastructure.
None of that is the kind of feature a commercial version would bother with. There’s no growth reason to tell people your data is 66% precise instead of just showing a pin. There’s no reason to keep dead sculptures on the map. Both cost effort and buy nothing except being right, which is exactly the kind of thing that survives when you’re building for an audience of one.
Why bother
I don’t think ad-supported reference apps are a moral failing. Someone has to pay the hosting bill (I host it on Cloudflare for free), and ads are usually the way that happens. But once the only user you’re accountable to is yourself, the whole calculation disappears. There’s no reason to hold a page hostage behind a banner, no reason to soften an inconvenient number, no reason to keep a feature that exists to extend session time instead of answer the question. Hyperagent and Claude Code made the cost of building these low enough that “just make the honest version” stopped being a hypothetical and became a Saturday (and Sunday).