This is the first post on this site, and I want to start with the problem I keep coming back to: public information that is legally available and practically unreachable.
Most government data is "public" the way a spare key is public when it's buried six inches underground in an unmarked lot.
You may have the right to it. You may even have a statute that says so. But if accessing it requires knowing which office holds it, which portal replaced the last portal, which PDF contains the attachment, which agenda item authorized the renewal, and which staff member understands the difference between a purchase order and a contract amendment, then the "public" part is mostly theoretical.
I spent years inside local government records. Later, I taught students whose families were trying to navigate court filings, immigration paperwork, housing problems, school systems, and benefits forms from the outside. The same pattern kept repeating: the information technically existed, but the path to it was institutional knowledge.
Institutional knowledge is a tax.
It is paid in time, confidence, phone calls, wrong forms, missed deadlines, and the quiet shame of feeling stupid in front of a system that was never designed to explain itself. Professionals do not notice the tax because they amortize it over years. Everyone else pays it all at once.
This is why I am increasingly skeptical of civic technology that treats "open data" as the finish line.
Publishing a dataset is good. Publishing a portal is good. Releasing PDFs in response to a public-records request is good. None of those things, by themselves, make the data usable. A file is not an interface. A portal is not a mental model. A scan of a contract is not public accountability.
The missing layer is not ideological. It is operational.
Can a resident search by vendor name?
Can a reporter find every contract over $100,000?
Can a council member see which agreements renew automatically in the next 90 days?
Can a local observer tell whether an item on Tuesday's agenda is a new purchase, a renewal, a sole-source award, or the fourth amendment to something originally approved five years ago?
If the answer is no, the data is still buried.
The strange thing about 2026 is that the technical part is no longer the hard part. OCR is cheap. Embeddings are cheap. Search is cheap. Extracting vendor names, dates, dollar amounts, and renewal clauses from ugly PDFs is not trivial, but it is no longer exotic. A small team, or one stubborn person with good tools, can now build interfaces over records that used to require a newsroom, a legal department, or a consulting budget.
That should change our expectations.
For a long time, governments could plausibly say: "The records are public; making them usable is too expensive." That argument gets weaker every month. The cost of turning static records into searchable public infrastructure has fallen dramatically. What remains is not primarily a technology constraint. It is a priority constraint.
This matters most in boring places.
Not the big national datasets. Not the dramatic leaks. Not the dashboards that win awards. The real leverage is in ordinary local documents: procurement contracts, agenda packets, budget amendments, vendor renewals, staff memos, redaction logs, invoices, service agreements.
The boring documents are where the governing happens.
They are also where opacity compounds. A software contract renews quietly. A vendor becomes the default. A temporary system becomes permanent. A price increase is treated as administrative. A scope change gets approved because no one in the room has the previous version in front of them.
None of that requires corruption. It only requires fragmentation.
Fragmentation is enough.