The aggregate table and the anonymized profile-level dataset are published here so readers can check the ranking instead of trusting a summary.
Downloads
| File |
Rows |
Format |
What it contains |
| results.csv |
16 |
CSV |
One row per browser with the published aggregate metrics. |
| results.json |
16 |
JSON |
The same aggregate data plus study metadata. |
| profile-results.csv |
12,800 |
CSV |
Anonymized profile-level check counts, status, launch time, and RAM. |
| checks.json |
47 |
JSON |
Check IDs, categories, and pass criteria. |
| README.md |
— |
Markdown |
Schema, definitions, and known limitations. |
Aggregate schema
| Column |
Meaning |
check_level_pass_pct |
Share of all 47 checks passed across all 800 profiles. |
full_profile_pass_pct |
Share of profiles that passed all 47 checks at once. |
avg_checks_passed |
Average checks passed per profile out of 47. |
facebook_ban_rate_90d_pct |
Descriptive 90-day ban rate, 400 Facebook profiles. |
instagram_ban_rate_90d_pct |
Descriptive 90-day ban rate, 400 Instagram profiles. |
combined_ban_rate_90d_pct |
Unweighted average of the Facebook and Instagram rates. |
survival_90d_pct |
Active and unrestricted at day 90; includes bans, checkpoints, and other losses in the denominator. |
median_launch_ms |
Median cold-start launch time on the test rig. |
ram_per_profile_mb |
Resident memory per profile after first page load. |
max_stable_concurrent |
Highest stable simultaneous profile count on the rig. |
Profile-level schema
| Column |
Meaning |
profile_id |
Anonymized profile identifier. |
browser |
Browser under test. |
platform |
Facebook or Instagram. |
proxy_country |
NL, DE, or US. |
checks_passed |
Number of the 47 checks passed for this profile. |
full_pass |
1 if all 47 checks passed at once, otherwise 0. |
failed_checks |
Up to three check IDs that failed; empty for a full pass. |
launch_ms |
Cold-start launch time for this profile. |
ram_mb |
Resident memory for this profile. |
status |
active, banned, checkpoint, or other_loss. |
ban_90d |
1 if the account was banned during the 90-day window. |
Notes
- The profile-level file contains no account credentials, cookies, personal data, or platform identifiers.
- Counts are rounded where the raw measurement had more precision than the published metric.
- The combined ban rate is an average, not a pooled count. The survival rate is separate from the ban rate.
- The aggregate table is the published result. If a row-level calculation differs by 0.1 point because of rounding, the aggregate table is authoritative.
Questions about the data: research@antidetectdata.com.
