Where the zero comes from
It counts software behaviour rather than an intention. There is no scheduled job here, no queue of addresses waiting to be tried, no client that speaks to Tor and no code path that opens a connection to anything. A request arrives and a page goes out. Nothing else happens on the way.
So the zero is counted in the strict sense used on the counted page. It is not an estimate of how often checking happens. It is the absence of the machinery that would make checking possible at all.
Why the zero is deliberate
A check from one machine, on one network, at one moment, tells you very little. Onion services fail in ways that are local, partial and brief. A circuit fails to build, an introduction point is loaded, a client sits on a bad path. Any of those produces a failure that has nothing to do with whether the service is running, and a success proves only that one connection worked once from one place.
A stale result is worse than no result. A green tick with a timestamp beside it gets read as current whatever the timestamp says, and people act on what they read. Somebody who sees no status makes their own attempt and learns the truth in the same twenty seconds. Somebody who sees a tick from two hours ago and cannot connect goes hunting for a replacement address, which is the exact moment forgeries get accepted.
And a site that probes builds a record of what it probed and when. That log exists whether or not anybody wanted it, on a host somebody else administers, subject to whatever eventually happens to that host. The safest log is the one nobody wrote, and the way not to write it is not to run the check.
What the zero costs you
This is a real cost and it deserves stating without softening. If you came here to find out whether an address is answering right now, you will not find out. There is nothing to consult and nothing to refresh. You will have to try it, and if it fails you will have to try another, and this site will not narrow the search for you.
A second cost lands on the reader who is already anxious. The moment somebody most wants a status board is the moment an order is late and an address will not load. Every page here is written for the calm case and will feel useless in the panicked one. That follows from the decision rather than being an oversight in it.
What comes back in exchange is that nothing published here can be wrong about the present. These pages describe formats, arithmetic and definitions, none of which change between the writing and the reading. A page making no claim about right now cannot go stale about right now, which is why a length is a safer thing to publish than a status.
What people read into the zero
The first conclusion people draw is that a site with no status board has been abandoned. Absence of live data reads as neglect, because most sites that stop being maintained stop changing. Here the absence is the product. The pages change when the text needs editing and never on a schedule designed to look busy.
The second is that a site showing a status is better informed. It is more confident. Confidence and knowledge come apart quickly in this subject. A status board is usually one machine running one check on a timer, publishing a result its own operator would not plan around, and the polish of the presentation says nothing about the quality of the measurement underneath.
The third is subtler. People assume that a site declining to check must think checking is worthless. Checking is not worthless. Your own attempt, from your own client, at the moment you care about, is the only check whose result applies to you. The argument here is against publishing somebody else's check as though it applied to everybody.
What would move the zero
Nothing planned. The decision is not a resource limit and more equipment would not reverse it. Adding a check would add the log, the staleness and the false authority as a single package, and there is no version of that trade that improves the site.
Wording can change. If a page here ever implies a claim about availability, that is a defect and it gets rewritten. The rule is stated at the foot of every page and again on the page about reading a page.
What will not change is the separation underneath it. An address that opens and an address that is genuine are two different claims, and only the second one has a real answer. That answer is on the forty page and it does not require anybody to run a check.
Questions people ask
So how do I tell whether an address is working?
Try it in Tor Browser. That is the only check whose answer applies to you, at the moment you care, from the network you are actually on.
Would you add a status page if it could be made accurate?
Accuracy is the part that cannot be fixed. One vantage point cannot describe a network that fails locally and briefly, and the log those checks produce would still exist afterwards.
Does this site record anything about me?
There are no forms, no accounts, no cookies and no tracking on these pages. The reasoning that keeps the checks at zero applies to everything else the site could have collected.