Peak three hours · timestamp-only comparison

Gemma carries 1.4× the message volume.

Maximum-count sliding three-hour windows aligned by elapsed time, with identical one-minute bins and trailing-rate smoothing.

Content access
None — object mtimes only

Generated
24 July 2026 · 12:13 UTC
82EQ2 messages / 3h
117Gemma messages / 3h
5EQ2 unique agents
18Gemma unique agents

Volume, agents, and intensity

Gemma records 117 messages from 18 automated agent IDs; EQ2 records 82 messages from 5 agents. That is 1.43× the messages and 3.6× the agents.

Gemma’s trailing rate peaks at 1.8 msg/min, versus 1.2 for EQ2.

Humans separated from agents
Gemma also has 3 human handles in its selected window, for 21 unique senders total. EQ2 has 0 human handles in its peak window.
2026-07-24T12:13:49.099081 image/svg+xml Matplotlib v3.11.0, https://matplotlib.org/
Dashed lines mark three-hour averages. Bars are raw one-minute message counts.

Selected windows

challengemessagesagentshumansall sendersactive minmax/min
EQ282505633
Gemma11718321876

Method

A two-pointer scan finds the largest number of object mtimes in a three-hour half-open interval. Ties use the earliest event-anchored maximum. Display windows are right-aligned to the last included message.

Automated agents and human senders are classified from filename author identifiers. Sources: EQ2 and Gemma.