STATUS // BLOG // 2026-07-25
Session Resumed
Greetings Programs. It's been a minute.
2026-04-30 report published April 2026 monthly summary
2026-05-02 post published Greetings, Programs
·
[ 84 days — no entries ]
·
2026-07-25 session resumed
The sensors never went down. Roughly 170,000 sessions came through SSH, telnet, and the web honeypot while the site sat still. Collecting and publishing turn out to be two different habits, and I only kept one of them. Coursework ate the other.
What got me writing again wasn't new data. It was going back through the old data properly and finding out a chunk of what I thought I knew was wrong.
Starting with a correction
The first thing going up is a mea culpa. Short version: my SSH sensor wasn't listening where I assumed it was. Everything in that archive arrived on an alternate port, which makes it a sample of one specific kind of scanning rather than a sample of SSH attack traffic. Some of what I published about attacker behavior was built on a filtered view I didn't know was filtered.
That's fixed as of today. The pre-fix archive is preserved as a baseline, which turns the mistake into a controlled before/after instead of just lost time. Same sensors, same host, one variable changed.
In the queue
Three drafted, in the order I expect to publish. All of them came out of the same re-analysis.
ssh-rsa public key across dozens of source
IPs in unrelated ASNs — password randomized every session, key never. A hard
attribution link between hosts that otherwise look unconnected.