Let’s consider two possibilities:
1. Supports do not contribute as much to the team’s success as core players.
This was the argument of CRAP, who drafted a team this season that 0-2’d Cerys, before mercifully being forced to disband by great adminship.
While I agree with his idea in general, I think it was presented in a pop-science way. Like when you read an article by a journalist, paraphrasing an “expert” for the hoi polloi, about a topic in which you specialize.
The belief that core players contribute more is also already held by basically everyone, except a few. So CRAP’s theory was akin to telling everyone what they already think, to critical acclaim.
___
But why do support players not contribute as much?
If matchmaking works, and if it only uses mmr, then shouldn’t a support player’s mmr reach a number that corresponds to his game-winning ability, just as much as a carry player’s mmr does? (Or does the matchmaker differentiate between core and support mmrs when balancing?)
In other words, shouldn’t the lower-contribution of supports already be “baked in” to the mmr?
___
By the way, by CRAP’s formula, you would put all your mmr in the mid-lane, if you were able. Indeed, he says:
“In auction draft, your single biggest goal should be to fill those top 2 positions with the highest performing players you can.”
Vel-polo put that theory to the test this season, and Polo5885 was the only high mmr captain to not make playoffs, as ForTheSwarm noted.
2. A second idea, which is much more interesting to discuss, is that the ideal mmr distribution on the team puts more mmr in core roles. —
- Not because mmr is not accurately reflecting winning-ability (which I mildly disagree with),
- and not because core players contribute more (which I agree with),
- but because mmr in core is better for strategic reasons.
Let’s briefly examine, and the reader can go into detail as an exercise on his own.
We’ll start with with two teams, and all the players have 5000 mmr. Every lane is a stalemate.
Let’s say, on a fundamental, theoretical level, that our goal in Dota is to win lanes. Let’s try to distribute our mmr to accomplish that.
Scenario 1: We’ll take 1000 mmr from the offlane, and give 500 to our mid, and 500 to our safelane.
Now, our offlane loses, 9000 mmr vs. 10,000. Our midlane wins, 5500 vs 5000. Our safelane wins, 10,500 vs 10,000.
Both teams still have the same amount of mmr, but now we win the lanes.
Scenario 2: We’ll again start with everyone having 5000 mmr. Let’s say that we want to help our mid player win his lane, because we correctly think mid has an outsize impact on the game.
Crudely, let’s take a 5000 mmr player away from the safelane, and he’ll secure the rune and gank mid.
Now our safelane is losing (5000 vs. 10,000), but our midlane is winning (10,000 vs. 5000). If the losses and gains were equal, we’d say it’s a good trade-off, because mid is more important.
But perhaps, to swing a lane, 5000 mmr is overkill and weakens the abandoned lane too much. Let’s say 3000 mmr is sufficient.
Now our safelane is (7000 on the carry vs. 10,000), and the roamed midlane is (8000 vs. 5000).
Scenario 3: Starting from the end of Scenario 2, let’s say the opposing offlane matches our idea. They will send a 2500 mmr support to roam.
Now, our safelane is (7000 vs. 7500), losing. Our roamed midlane is (5000 + 3000 vs. 5000 + 2500), perhaps with an edge.
___
For brevity, I’ve limited myself to a few examples. I hope you found it stimulating.
For my friends, a short autumn playlist: https://www.youtube.com/playlist?list=PL41FGYc9Pm4eI5v_RB7vwXDCbpStuNcjb
Public Notice 