The Catalog Is Not the Record: A 2026 Freshness and Completeness Audit of Illinois' Open Data Portal
Illinois maintains an open data portal, and that fact alone is often treated as proof of transparency. A catalog of datasets is a useful starting point, but it is not the same thing as a public record. The relevant questions are narrower and more demanding: Is the data current? Is it complete enough to check a claim? Can a resident actually use it without specialized help? And what still requires a Freedom of Information Act request?
This audit is an editorial review, not an official certification. It examines the portal the way an engaged citizen would: by reading metadata, sampling datasets, checking update claims against record dates, and comparing what agencies publish proactively with the records residents most often have to request. The findings are not a indictment of the portal's existence. They are a reminder that open data is a records program, not a website feature.
How We Audited the Portal
The review focused on three tests. First, freshness: does the dataset's stated update date match the period covered by the records inside it? Second, completeness: does the published file contain the level of detail needed for accountability, or only an aggregate summary? Third, usability: is the data machine-readable, documented, and stable enough to download, sort, and compare over time?
Those tests matter because portals often fail quietly. A page can look maintained while the underlying records are months or years old. A dataset can be technically public while its field names are unexplained, its identifiers inconsistent, and its download links unstable. The result is a familiar Illinois pattern: the portal advertises openness, while residents still file requests to get the current version of the record they need.
Freshness Is a Claim, Not a Fact
Many catalog entries include an updated date, but that date can mean almost anything. It may reflect when a file was uploaded, when a webpage was edited, or when an agency ran a routine export. It does not necessarily reflect the most recent record in the dataset. A budget dataset updated annually is not stale merely because it is old; it is stale if the agency presents it as current without saying which fiscal year it covers. A contracts dataset is stale if the newest row is from a prior quarter while the page carries a recent update stamp.
Agencies should publish a plain freshness statement for each dataset: what period the records cover, how often the dataset is refreshed, when the next refresh is expected, and who owns the data. Without that statement, citizens are left to guess. The practical fix is not expensive. It is a metadata standard, applied consistently, with record-period dates separated from page-update dates.
Aggregates Crowd Out Transactions
The second gap is completeness. Open data portals often publish summaries because summaries are easier to produce and less likely to raise follow-up questions. But accountability usually lives at the transaction level. A resident cannot verify a vendor payment pattern from a category total. A reporter cannot check a contract amendment from a summary table. A watchdog cannot trace a grant from appropriation to recipient without row-level records and stable identifiers.
Not every record can or should be published immediately. Privacy, security, and legal restrictions are real. But the default should favor proactive release of the records residents most often request. If an agency repeatedly receives FOIA requests for the same contract list, inspection log, or payment file, that repetition is evidence that the portal is missing a public dataset. The FOIA residual—what remains behind request—is one of the best measures of whether open data is working.
Usability Is Part of Openness
A dataset that cannot be understood is only formally open. PDFs and scanned images are common on government sites, but they are poor substitutes for structured data. So are spreadsheets with merged cells, unexplained abbreviations, inconsistent agency names, and no data dictionary. If a resident cannot join a spending file to a contract file because the identifiers do not match, the portal has published two documents but not one usable record.
Usability also means stability. Links should not break when an agency redesigns a page. Bulk downloads and application programming interfaces should be documented. Historical versions should remain available so that comparisons are possible. These are basic records-management practices, not technical luxuries.
A Citizen Checklist for Spotting Stale or Missing Data
- Compare the page update date with the newest date inside the records. If they diverge, ask which one is accurate.
- Download a sample and sort by date. The most recent row should match the stated coverage period.
- Look for a data dictionary. If field names are unexplained, the dataset is not fully usable.
- Check for stable identifiers. Agency names and vendor names may vary; unique IDs should not.
- Confirm whether the file is an aggregate or a transaction-level record.
- Compare the portal dataset with the agency's budget, contract, or payment documents. Gaps often appear at the edges.
- If you file a FOIA request for records that should be public, ask the agency to add the dataset to its publication schedule.
What Agencies Should Do Next
The strongest reform is also the simplest: treat the portal as part of the agency's records program, not its communications shop. Publish a freshness standard. Require data dictionaries. Prioritize transaction-level records where law and privacy allow. Keep a public list of datasets scheduled for release, and explain delays. Track FOIA requests by subject and use repeat requests as a signal for proactive publication.
The measure of an open data portal is not the number of catalog entries. It is whether residents can answer basic questions without filing a request and waiting. Illinois does not need a grand new promise in 2026. It needs agencies to publish what they already produce, document it clearly, and keep it current. A catalog is not the record. The record is what the public can actually see, use, and verify.