The Registry's own defect records for Special Leave Petitions, read in full.
Before writing product code, Nyave manually analysed the recorded defects across more than 2,000* Supreme Court Special Leave Petitions. The figures below describe Nyave's own dataset and nothing beyond it.
of the filings in Nyave's dataset carried at least one defect at first submission.
The rate held between 96.9% and 98.6%* in every one of the eight years.
The pattern does not show that AORs are careless, that the Registry is slow, that every defect causes the same delay, or that every recorded defect could have been prevented. It shows that recorded defects were recurring rather than isolated within the analysed set.
Share of each year's filings with at least one recorded defect, in per cent.*
A median of six defects. One filing carried 46*.
A mean of 6.8* defects per defective filing; one in twenty carried 14* or more; one carried 46*.
The Registry's defect sheet arrives about two days* after filing.
of defective filings needed more than one round of cure; one ran to eight.
The single most-cited rule in the dataset: about a quarter* of all recorded defects, near zero before 2025, 45%* of 2025's and 60%* of 2026's.
filings in the dataset were dismissed for non-refiling after 90 days.
Seven things appear in no public Court source and exist only in this dataset, among them:
How the dataset was built and checked.
The Registry's own public defect records, 2019 to 2026.
The dataset holds the defects the Registry recorded for each filing. It does not hold the petitions.
Every record was reviewed manually and cross-checked by program.
A gap the cross-check found was corrected across every filing. The corrected dataset is the one used.
A separate sample of 30* Special Leave Petitions each from 2024, 2025 and 2026 found 99%* defective.
The engine itself is tested on real filings: a library of 12* real Special Leave Petition filings, with their actual Registry defect sheets, is the ground truth.
TechnologyTwo limits should be read with those figures.
First, the dataset is Nyave's own, assembled for the purpose of building a verification tool; it is not a random sample of all Special Leave Petitions filed in the period and it is not an official estimate of how often Special Leave Petitions are found defective.
Second, "recorded defect" means a defect the Registry noted at scrutiny, as captured in the dataset; a count of recorded defects does not weigh how serious any one of them was.
* Figures are rounded and based on Nyave's own dataset. The dataset is not a random sample of all Special Leave Petitions and should not be read as an official prevalence estimate.
Each of the 106* rules carries the outcome the Registry attaches to it, measured on these records.
The dataset confirms how often. The official sources confirm what is correct.
