I agree that MM for DB is more difficult than Quake (or chess, where the Elo rating originates from) but I still believe there’s approaches that could yield better results than the current one. The MM should attempt to measure how capable players are in different roles (engineer, medic, combat), their general playing style (aggressive/defensive), there willingness to adjust (e.g. switch mercs if needed) and so on. With these data the MM can compose teams that have all the positions covered.
There also would be ways to measure the skill level in a role using slightly more creative KPI’s than combat/objective/support XP, e.g. distance moved while defusing or repairing, health charge when reviving, ratio of reload cancel etc.
Hopefully the new MM moves in a direction like that.