Unwanted e-mail and slow thinking

Last year I mentioned that I do not graylist incoming e-mail because of how some senders retry from different hosts but I do use IP blacklists to block MTAs. This approach has been quite effective though in recent weeks more spam in East Asian languages has been getting through and brought my anti-spam arrangements back to mind.

I am intelligent and, about some technical things, quite knowledgeable. Despite this, sometimes it takes me a long time to see the obvious. The example in this case is the synergy between graylisting and blacklisting: the very simple point that I could well receive much less spam if I graylist for long enough for senders to be added to the blacklists for which I fetch hourly updates.

Such a use of graylisting does not require the typical specificity about the envelope sender and recipient: I do not care who e-mails whom, I care more that the MTA is attempting a delivery to anybody at all. I wonder if there is a configuration option for that. If not, graylisting by sending IP only may be simple enough that I can arrange it between pfctl -t and cron without a fancy dæmon if a failure to connect is taken rather like a 4xx SMTP response.

Facial recognition and witness protection

Facial recognition algorithms seem to be becoming rather good. Not being a Facebook user I have been distant from DeepFace but British police forces have been trialling facial recognition and I have noticed various news of successes in automated identification of criminals from cameras in cities like Cardiff.

Between the increasing ubiquity of both private and government cameras and the rate at which photographs and videos are posted to social media, I wonder if we are seeing the end of practical long-term witness protection. After all, big data is a hot topic and reflectacles go only so far.

I should note that here in Scotland the term for such witnesses is protected persons, not to be confused with the British immigration term.