The above calculations do not explain why we are seeing some instances of oddness where a few lines have had DLM action taken at a ludicrously low rate of errors.
In other news I can however confirm that Openreach are playing and looking at employing a temporal DLM which has been up for trials since Sept last year. This info is in the
public domain so I am not giving away trade secrets.
Temporal DLMs do not run for a 24hr period and may only record data during either certain times of the day or may run for differing time periods. (with the same params). The idea of temporal DLMs is to increase stability for lines whose 24hr period MTBE is in effect negated by large parts of the rest of the day without errors.
Please note it is very important that this is/was a smallish trial and
not in general use yet. Only a tiny portion of lines are on the trial and that is only
if your ISP is one of those elected to go on the trial.