[New] Autoscramble Issues Evidence Thread

[NovaCancy]

Hey y'all!

There have been a lot of concerns regarding how autoscramble functions and how sensitive it is, with multiple complaints that it often doesn't occur soon enough. We hear and understand your concerns, and would like any evidence (usually screenshots of the leaderboard work) submitted here with a detailed description of how the game went, if autoscramble ever occurred at all, what map it was on, and any other information you think might be relevant.

It is also worth noting that this thread is not for scrambles themselves, autobalance, votescrambles, or how effective a scramble is. This is just for issues regarding when the autoscramble plugin detects that the teams are unbalanced and scrambles them without a vote or admin forcing it.

Thank you for your help!

 

[Izaura]

Map: pl_badwater
Server: EU Standard
Date: 29-March-2024, around 8:20PM (UTC +1:00, CET)
As Blue: Did not manage to cap the very first point, ran out of time.
As Red: Unable to defend properly, steamrolled all the way to last.
Final score: 8-0

Scramble only triggered after the second round, when the map was already over.
Not enough people voted for a scramble.
Screenshot attached, leaderboard at the very end of the map.

 

[Izaura]

Map: cp_steel
Server: EU Standard 2
Date: 29-March-2024, around 10:30-10:40PM (UTC +1:00, CET)
As Blue: Red did not manage to defend points, managed to cap A through E in order and in quick succession.
(I've tried the funny strat where you setup as engi on B, but failed. It didn't matter though as my team capped B immediately, i did nothing lol)
As Red: Blue didn't manage to cap A in time, so they tried the funny E cap but failed, no points capped.
Partial Score: 5-0

Scramble didin't trigger after the second lost round, players eventually voted for a scramble.
Screenshot attached *should* be leaderboard before scramble, but i sadly lost track.

 

[Izaura]

Note: I'm changing date format to YYYY/MM/GG, i would change it to the previous posts as well, but i can no longer edit them.
Also keep in mind, due to daylight savings, date might be an hour off.

Map: pl_barnblitz
Server: EU Standard 2
Date: 2024-March-31, around 3:35PM - 4:00 PM (UTC +1:00, CET)

Round 1, as Red: Blue couldn't manage to push the first point.
Round 2, as Blue: Red couldn't hold their defense for long, last wasn't hard to push either.

• At this point, only 4 people voted for a scramble, so teams remained unscrambled.
• Plugin didn't trigger a scramble.

Round 3, as Red: Went slightly better for blue, as they managed to push the first point.
However they ended up pushing during overtime and failed to push the second point.

After round 3, the plugin DID trigger a scramble, but map was already over.
Final score: 1-4

Attached three screenshots:

1. The leaderboard after the second round,
2. The leaderboard after the third round,
3. A screenshot of the plugin triggering a scramble, but way too late as the map was changing.
   (I think I've cut it off a bit, as i was hurrying up before the map changed).

I would also like to draw attention to the domination count of both teams.

 

[NovaCancy]

Round 1 of Upward, no scramble was initiated.

 

[VintagePC]

A couple of comments that came out of a discussion on this today:

The scramble thresholds are currently:

• 10 more dominations,
• 2x the frag ratio, or
• 3 consecutive full-round wins.

Part of the challenge here is that game modes and maps break this down differently as to what is "a round", and that means I don't think there is a good universal metric. One would have to write a far more involved plugin that handles everything differently depending on whether it's payload/AD/CTF, etc. I don't have the SM skills or game logic knowledge to do that, and other scramble plugins that I'm aware of are not TF2C compatible or can't handle 4 teams.

- Keep in mind the scoreboard is not an accurate representation of frags, since you can a lot of points for doing things that are not kills, so tracking frags is not straightforward from the information here.

So far everything posted above seems consistent with the behaviour as configured. e.g. for cp_steel, there was an incorrect expectation of a scramble after round 2, and in other cases the maps have long full rounds due to a number of mini-rounds or stages.

We're looking to mine some of our logs to see what are "typical' values for those thresholds and if we can refine the configured limits a bit more.