# Any reason voting can’t be a subsystem?

**URL:** <https://forum.thousandbrains.org/t/any-reason-voting-can-t-be-a-subsystem/1201>\
**Category:** General\
**Created:** [September 18, 2026, 11:12pm UTC](https://forum.thousandbrains.org/t/any-reason-voting-can-t-be-a-subsystem/1201 "2026-09-18T23:12:12Z")\
**Posts on this page:** 3\
**Page:** 1

<div class="post-metadata">

**Author:** ![evanuno](https://yyz2.discourse-cdn.com/flex034/user_avatar/forum.thousandbrains.org/evanuno/32/248_2.png) [@evanuno](https://forum.thousandbrains.org/u/evanuno)\
**Post date:** [September 18, 2026, 11:12pm UTC](https://forum.thousandbrains.org/t/any-reason-voting-can-t-be-a-subsystem/1201/1 "2026-09-18T23:12:12Z")

</div>

It seems like a subsystem that listens for votes from all columns and reports the results back to all is the most scalable solution for voting. If it floods the network once there’s a big cluster with 1000 or more instances of Monty all voting at once, a network interface just for voting could be added to the hardware. Multiple votes could be managed simultaneously because the votes would include a tag. Something all the columns know. No intelligence would be required, just math. An id for the column could be included for telemetry but not used otherwise. Things that would be useful in the voting results reported to the columns include % voted, and for, against, or don’t know. It doesn’t need numbers of votes. Not voting is not counted because it won’t know how many columns there are or how many should be voting at that time. A telemetry processor for that system could reveal that. On a single system running multiple columns it could be done with sockets for maximum speed. Multiple independent Monty systems could use the voting system at that level to identify extremely complex objects.

Everything the brain creates has clues to the way the brain works internally, and voting is something the brain has already solved multiple times. Localized voting gives you ant colonies, herds and flocks. The human brain is able to use an external system to process the voting and return the results. We can only do that because of language, and we only have language because it is based on how the brain communicates internally. It follows that we vote the way we do because our brain votes that way. It has no other base to build on.

---

<div class="post-metadata">

**Author:** ![tslominski](https://yyz2.discourse-cdn.com/flex034/user_avatar/forum.thousandbrains.org/tslominski/32/5_2.png) [@tslominski](https://forum.thousandbrains.org/u/tslominski)\
**Post date:** [September 18, 2026, 11:21pm UTC](https://forum.thousandbrains.org/t/any-reason-voting-can-t-be-a-subsystem/1201/2 "2026-09-18T23:21:23Z")

</div>

> [@evanuno](#):
>
> Things that would be useful in the voting results reported to the columns include % voted, and for, against, or don’t know.

Hi @evanuno, one highlight I want to make is that in the current implementation of Monty, the votes do not map into for/against in the typical sense. [Votes contain “information about possible objects and their poses”.](https://docs.thousandbrains.org/docs/evidence-based-learning-module#voting-with-evidence)

---

<div class="post-metadata">

**Author:** ![evanuno](https://yyz2.discourse-cdn.com/flex034/user_avatar/forum.thousandbrains.org/evanuno/32/248_2.png) [@evanuno](https://forum.thousandbrains.org/u/evanuno)\
**Post date:** [September 20, 2026, 4:40pm UTC](https://forum.thousandbrains.org/t/any-reason-voting-can-t-be-a-subsystem/1201/3 "2026-09-20T16:40:33Z")

</div>

Thanks, does that prevent a subsystem from assisting?