On 28 Jul 2026, at 08:54, James Bensley <james@inter.link> wrote:
Hi Jeroen,
Nice work! Thanks for doing this and sharing your findings. From looking through the list archives, I feel like the last time around in 2021 this kind of homework wasn't done and this really makes it clear why this approach doesn't work with empirical data. Thank you.
The above is just the RIPE data, adding APNIC/ARIN/AFNIC/ etc would likely explode the data even more and find more prefixes. Do remember that for forever has existed: https://www.cidr-report.org/#Bogons Which is mailed every month or so to the various NOG lists. The problem here is transits who are not verifying the prefixes coming from their 'customers'. And as the time shows, nobody ever bothered to filter those out, bits is bits when you forward them and get paid for them, be that good or bad data.... Thus knowing what is bad is known, cleaning that up.... at least with converting the data one could stop it at your network edge, but the bits might still come to you. Regards, Jeroen