How sources are found and checked

What's in the directory

The directory has an entry for each of Ohio's 88 counties. It has one for each of the 87 cities with 20,000 people or more in the 2020 Census.

For each county and city, the entry says where three kinds of records are posted: bids, agendas and commercial building permits. If a place doesn't post one of them online, the entry says what to do instead.

The state's agencies add their own bids, business filings, liquor permits, professional licenses and public notices. Local health districts add food-service inspections. The Federal Motor Carrier Safety Administration adds Ohio's trucking companies, bus companies and freight brokers.

Villages, townships and smaller cities don't have their own entries. Some of their records are kept by their county.

Records about private people are left out: permits for homes, property sales, court cases, foreclosures, probate and name changes.

How a source is found

Each source was found on the government's own website.

The entry records the page's address and the system it runs on. It records how the records can be reached, how often new ones appear, and the site's terms of use, quoted.

Each page was opened once to confirm its address. The date is on the source page.

Some sites offer their records in a form made for programs, such as a data service or a download file. Where they do, the checker reads that instead of the page people use.

When one office keeps another's records, the entry points there. A county may keep the building permits for its townships, for example.

The first entries were gathered in November 2026.

They were gathered with AI research tools, under Frankie Cleary's direction.

How each source is checked

Once a night, the checker opens the page or file where each source posts its newest records.

Where the site supports it, the checker first asks whether anything has changed since the last visit. If nothing has, the site says so in a short reply, and that check is done.

If something has changed, the checker reads the newest records. It stops at the first one it has seen before.

It checks that the records could be read. It also checks that they carry the same kinds of details as before, such as a permit number and a date.

Then it sets the source's status. Each status follows a fixed rule, the same for every source.

In its first nights on a source, the checker also reads the source's older records, a little each night, at the same slow pace. After that, a night costs each source one or two requests.

A private copy of each page and file the checker reads is kept, exactly as published, so nothing has to be read twice. The copies are never shown or sold.

Each source page shows its last 30 checks.

Sources updated by hand

Some sites can't be read by an automated checker, but a person can download the records from the site's own pages.

For some of these, a person downloads the records on a set schedule: every week, every month or every three months. Some of those sites need a sign-in, and the person uses their own account. The checker reads the copy they download.

The source page shows the schedule and the date of the last update, as in "Updated by hand monthly, last on October 3, 2026."

When an update is later than its schedule, the status says Update due.

What each status means

Working: The last check read this source. Its records came through the way they did before. Use the link. It worked at the time shown.
Changed: The last check read this source, but something about its records has changed. The reason is shown beside it. Use the link, and check the dates on the office's own site.
Not working: The last check couldn't open this source, or couldn't read the records on it. Try the link. If it fails for you too, try the office's main website or call the office.
Blocked: The site asks automated checkers to stay off this page, or it turned the checker away. The checker stopped there. Use the link. The page may work in your browser.
Not read yet: This source is in the directory and will be read, but the checker can't read this kind of page yet. Use the link.
Not checked: The night's check ran out of time before it reached this source. It doesn't hurry to fit every site in. Look at the check before this one, or use the link.
Not read: This source is in the directory, but the checker doesn't read it. The reason is shown beside it. Use the link, or the other way in shown beside it. You can also ask for it to be read.
Update due: A person updates this source by hand on a schedule. The last update is older than that schedule. Use the link for anything newer than the last update.
Site didn't answer: The site didn't answer within 30 seconds, or it answered with an error each time it was tried. Try the link later, or look at the next night's check.
Page moved: The page wasn't at its address, or the address now sends visitors to another site. Look for the page on the office's main website. If you find it, send the new address to frankie@newonrecord.com.
Layout changed: The page opened, but its records are laid out in a new way. The checker couldn't read them as before. Use the link. The page itself may be fine.
No records shown: The page opened but showed no records. The check before found some. Look at the office's site. The records may have moved to another page.
Nothing new: No new record has appeared in three times the usual gap between records. For a source that posts weekly, that's three weeks. Look at the office's site. It may be posting somewhere else now.
Kept out: The site keeps automated checkers off this page, by asking them to stay off or by turning them away. The reason is shown beside it. Use the link. The page may work in your browser.

A status says what the last check saw, and when. It isn't a promise about the next check.

Working means the checker read the source's newest records at the time shown. It doesn't mean every record the office keeps is online. Before you rely on a record, look at it on the office's own site.

What the checker never does

It never reads a page the site asks automated programs to stay off. Sites post those instructions in a file called robots.txt.
It never signs in. It never uses a password, a key or anyone's account.
It never gets around a refusal. When a site turns it away, or asks visitors to prove they're a person, the checker stops and the source shows as Blocked. It doesn't try again in a browser, under another name or from another address.
It never hides what it is. Every request carries its name and a way to reach the person who runs it.
It never hurries a site. It sends one request at a time to a site. It waits at least two seconds between them, and longer when the site asks.
It never reads a source whose terms of use don't allow it.
It never reads a source made up of records about private people. Where a source mixes some in, such as permits for homes, those records are dropped and never shown.

No page on this site shows records, or anyone's personal details.

How often

Each source the checker reads: once a night, starting at 2 a.m. Eastern time, or 3 a.m. during daylight saving time.
Each site's robots.txt file: read again once a day.
A source the checker reads that turned it away: checked again the next night, the same way. A site may turn checkers away one night and not the next.
A source marked as not read because its site said no: the page in its entry, on up to three nights soon after and then once a week, if the site allows that, to see whether the no still stands. No records are taken from it.
A source updated by hand: weekly, monthly or every three months, as its page says.
These pages show each check's results within an hour.

Something wrong?

If an entry is wrong, a page has moved, or a status looks off, write to frankie@newonrecord.com. Include the address of the page on this site.