My wife sent this site to a friend. I panicked, because it wasn’t ready. Then I did what the site tells everyone else to do: I stopped guessing and checked the record. Here is what the server saw, what I broke, what I took back, and what Google says it can see.
I call her my Sunflower. She has sent people here before — she did it the week of the State Fair debate. This time it landed differently, because I had been building for weeks with an audience of one, and suddenly the audience wasn’t me.
The honest first reaction was fear: somebody is going to click something that doesn’t work. The honest second reaction was the one this site is built on. Don’t argue with the feeling. Pull the record.
| What I believed | What the record said | |
|---|---|---|
| “The site isn’t ready.” | It has been live since September 2. A script compares every file on the server to my computer, byte for byte: 106 of 106 match. | WRONG |
| “I turned the security challenge off.” | I turned one off. A second, automatic layer still fires — and the only thing it caught was me. | HALF |
| “Google can’t see us.” | Google crawled the homepage on September 7. It is indexed. | WRONG |
| “The sign-up email bounces.” | It did on September 2. It doesn’t now. I sent a test from my phone and it arrived. | OUT OF DATE |
| “This is America’s first help desk.” | I checked that on September 9. It isn’t. It was still on 18 pages anyway. | FIXED TODAY |
Not one of the five held up exactly. Four were wrong in the direction of fear. The fifth was wrong in the direction of pride. None of them would have been caught by feeling harder about it.
My hosting company keeps a request log. I asked for a summary on the morning of September 11.
Here is who those requests came from, as the log summary described them:
I can’t tell you how many humans that is, and I’m not going to guess. One person opening one page makes a dozen requests — the page, the fonts, the data files. The only way to count people precisely is to track them, and the bottom of every page here says no tracking. I would rather publish “I don’t know” than break that to get a better number.
This site answers at two addresses — with and without the www. The tidy thing is to send one to the other, so on Thursday evening I asked my host’s assistant to set up that redirect.
Redirect created: plain address → www address.
The www address starts redirecting to itself. The homepage won’t load.
Redirect deleted.
Cache cleared so nobody kept getting the broken version.
Checked from outside: both addresses load. No loop.
My best guess at why: both addresses point at the same folder, so a rule set on one got applied to the other. That is a guess, and I am labeling it one. The fix I’m using instead can’t loop: leave both addresses working and let Google figure out which one is the real one, which it already does.
The week a stranger shows up is exactly the week you want to tidy up. That is also the week a tidy-up can break the front door. I am writing that down for next time.
I set the security level to Medium and turned off the browser challenge. The next morning, while checking that a new post had uploaded, my own computer got this page with a 403 — a “checking your browser” screen and an instruction telling search engines not to index it.
If that page were reaching Google, every page on this site could quietly drop out of search. So I asked the host directly. Their answer: it’s a separate, automatic attack protection that fires on bursts of fast requests, the threshold can’t be adjusted, and verified Googlebot gets through. I had made about ten requests in a few seconds. The guard was right to stop me.
“Verified Googlebot gets through” is the host’s word, not something I had seen. So I set up Google’s own tool, Search Console, to see what Google sees:
If the guard had stopped Google, that last crawl would have picked up the “don’t index” instruction and the homepage wouldn’t be there. It is there. That settles it for one page. The report that covers all 100 comes in a week or two, and I’ll post what it says either way.
On September 9 I asked to be called out on the word first, so we checked. It didn’t hold. Nebraska already has people doing parts of this, with training and pay — Nebraska Documenters sends residents to public meetings and publishes their notes, and there are nonprofit newsrooms doing data work. The page that explains how this site works opens with that failed check.
The footer on 18 other pages kept saying first anyway — including the headline on the front page. Today it came off all of them, 25 times, plus the four scripts that build some of those pages, so a rebuild can’t put it back.
A site that checks other people’s claims has to lose its own first. I’d rather be the help desk that took a word back than the one that kept it.
The fear said the site wasn’t ready. The record said it had been ready for nine days and I had been the one not looking. What actually needed fixing wasn’t what I was afraid of: one word that wasn’t true, one inbox I hadn’t tested, one folder my own checker skipped, and one redirect I shouldn’t have asked for.
Would you believe me if I told you the scariest part of having visitors is finding out they’re welcome? You don’t have to believe me. Everything above has a date on it.