Between November 2017 and September 2026, 93 federal district courts filed 693,779 cases under seal and 1,170,783 cases in public. Sealed filings are about 37.2% of the whole count. They are common. They are not the majority.
Of the sealed numbers, 89,435 — 12.9% — later appear as opened in this mirror. 604,344 do not. The year we have watched the longest, 2018, is still only 21.4% opened. So the headline is not an illusion of “we just haven’t waited.” Waiting raises the rate. It does not bring it anywhere near all of them.
The drop from 21.4% in 2018 to 5.4% in 2025 is partly the clock. Newer cases have had less time to be unsealed. 2026’s opened column is entirely blank in the sheet. That blank is missing follow-up, not a measured surge in secrecy. Anyone quoting 2026 as “nothing was unsealed” is ahead of the data.
The kind of number matters more than the year. In 2026 the project listed every sealed case number,64,491 of them. 72.7% are magistrate numbers. 8.1% are search warrants. 6.2% are criminal cases. A warrant is sealed so the search is not announced in advance. A magistrate complaint is often the first, quiet step before an indictment. Those are ordinary tools. They are not, by themselves, a secret trial.
From 2019 through 2025, sealed criminal numbers were later marked opened 71.4% of the time. All the other sealed numbers were marked opened 8.6% of the time. Read that with one mechanical fact: when a sealed magistrate matter turns into a public prosecution, it usually gets a new case number. This dataset follows the old number. A magistrate line can sit in the “still sealed” pile even after the defendant has been named in a public criminal case.
Courts are not interchangeable. Eastern District of Missouri opened 75.2% of 11,521 sealed filings. District of Massachusetts opened 2.2% of 26,553. Eastern District of Washington opened 1.9% of 8,201. They were counted over the same months. Magistrate share does not explain the gap — that relationship is essentially flat. Something in local practice, reporting, or follow-up does. This page cannot see which.
One reporting habit can fake a court’s whole rate. The Northern District of Illinois has a median month of 10 sealed cases, but 14 months are many times larger and hold 15,709 of its 16,615 sealed filings. Strip those rows out and the rest of its months opened 525 of 906. The national total barely notices — those flagged months are 2.4% of all sealed filings — but the court chart does.
Spring 2020 is the other trap. From March through May, public filings fell from 53,830 in 2019 to 23,202 in 2020. Sealed filings went from 19,463 to 18,092. If someone says secrecy spiked when the courts closed, they are looking at the share, not the count. The count of sealed filings held steady. The count of public filings did not.
And a named update is not a person. Of 28,677 captions matched to a public docket, 4,582 say search warrant. Phones, accounts, subpoenas, and pen registers make up another large slice. The case-updates page separates those before it lets you search a name.
The count sheet is the public workbook “Cases Filed Sealed 2026 [Mirror 1].” It says it can be viewed without permission and that it is not an official download feed. Historical months point at other mirrors when Google struggles to load.
Sealing is a judicial act with a public reason: an investigation, a witness, a warrant, a juvenile, a fair trial. The open question this mirror can actually answer is narrower. Of the case numbers that started sealed, how many does the same reporting later show as opened — and how different is that answer from one courthouse to the next?