As @anoek merged your PR, I assume he approved of the change.
Users should feel free to express their opinions, and then anoek can use his judgement. That’s how a benevolent dictator governance model works. 
I’ve problems to understand what I gain by looking on the “upsetting” results on the pie or the boxes below the rating graph.
There are plenty of games where I win against stronger and loose against weaker opponents. And the reason for this is simple: there is no category for even opponents. Either they get counted as weaker or stronger, even if they are almost the same rank as mine. And in even games I should win about every 2nd game.
What I could deduce by comparing the “expected” and “upsetting” ratio would be, if the player plays mostly even or if she plays opponents of different strengths.
To find the games with the really unexpected results (wins or loses against opponents with a rank difference of 2 or more) we would need a filterable history as GaJ suggested.
1 Like
Yes - Good point. A category for even opponents would also be useful.
The situation is even more complicated, because an opponent may be weaker at the start of the game and stronger at the end. The ratings at the end of the game are used to determine if the opponent is stronger or weaker.
Another point worth mentioning is that some users may be colour-blind. They may like to choose a colours to suit them.
Hi, I am a student currently studying about topics that involve UX and UI design, and from what I quickly glanced it seems that my current knowledge can be put in use here.
Can someone quick summarize for me the main points of what has been discussed so far, so that we can have a more productive discussion and quickly reach to a point were I can be truly useful into this discussion?
Thanks.
1 Like
(This summary will probably be tainted by my own biases.)
A few months ago, @GreenAsJade implemented a nice, beautiful pie chart for our wins/losses. When he introduced it, he swapped the previous meaning behind the shades of red used in the bar chart.
Previously, the colour code worked as follows:
Upsetting game results:
Intense shade of red = Losses against weaker opponents
Intense shade of green = Wins against stronger opponents
Expected game results:
Weak shade of red = Losses against stronger opponents
Weak shade of green = Wins against weaker opponents
Now the colour code works like this:
Games against stronger opponents:
Intense shade of red = Loss against stronger opponents
Intense shade of green = Win against stronger opponents
Games against weaker opponents:
Weak shade of red = Loss against weaker opponents
Weak shade of green = Win against weaker opponents
(Side note: I use “intense shade” and “weak shade” instead of “bright” and “dark” since the latter depend on whether you’re using the bright or dark theme for OGS.)
GreenAsJade’s argument is that upsetting vs. expected game results should not be a function of the bar and pie charts, but rather a function of the game history, which should be overhauled if @anoek isn’t needed for cooperation. GAJ argues that if the shades of red were swapped back to signifying upsetting/expected game results, people would not be able to readily compare their overall results against stronger vs. weaker opponents via the pie chart.
My argument is that upsetting/expected game results USED to be a function of the bar chart, whereas currently, they are not signified by anything. I also feel that if the shades of red were swapped back while the layout of the pie chart remained the same, “results against stronger” could still be readily compared against “results against weaker” since the former are displayed on the right side of the chart whereas the latter are displayed on the left side (and the chart has a nice caption explaining this, anyway). Furthermore, I don’t know how possible an overhaul of the games history is without anoek, whom we cannot currently rely on for cooperation.
At any rate, I think the issue is cosmetic. 
1 Like
I think this is a very fair summary.
I would like to add a little emphasis on what the pie chart is for.
Pie charts in general are a visualisation for ratios.
This one is designed for tracking two ratios: the ratio of win loss and the ratio of played vs weaker to played vs stronger.
These are important ratios for players to be able to track, because our ranking system works best when we all keep these roughly in balance.
In contrast, the bar graph is a bad tool for highlighting the amount and timing of unexpected results.
It did do that before, but so poorly that I didn’t even know it was used for that
That covers the rationale for the change.
What I currently think:
Players should be able to chose the colour scheme.
That would solve this debate, but more importantly would allow for colour blind and other issues.
What will happen next:
Likely nothing.
The current dev-review-release cycle is not conducive to changes like this, at least from my perspective, at present.
That means if it’s going to change, someone other than me would be doing it. That’s possible, but there aren’t any indications of it at the moment.
GaJ
I should add: it’d be a good first-ticket for someone who’s interested in hacking OGS client.
2 Likes
Well, in the end you are technically right. But you will see that things aren’t that simple actually. There isn’t much UX or UI design involved in this situation, of course. But the thing is, clearly there is some, so lets proceed with my observations about this.
At first, let’s make things clear to everyone using a quick image editing to visually represent the discussed topic. Correct me if I am wrong, but this is what is been discussed.
He is completely right in that assertion. A pie chart serves primarily the purpose of displaying ratios. The aforementioned feature would be better, from a UX perspective, as a filter in the game history.
Indeed. This statement here is pretty important, you will see why soon enough.
Sorry, but now you just contradicted yourself. You said that they don’t have any specific meaning, but just take a look on the your quote above this one, and especially, pay close attention to the specific way that you grouped the current color code. Yes, there is a meaning, it’s grouping together similar information, one can even that in the same way that it did before.
The concept behind everything in this discussion is a UI design one. It’s called Gestalt, specifically the Law of Similarity. This Law states a very simple purpose: Information that is grouped in a similar fashion will be interpreted by your brain as belonging into a similar group.
Again, pay close attention the way that you grouped the before and current color code. Before, you were used to group together in a upset/expected manner. Now, you are grouping into stronger/weaker manner. You applied the Law of Similarity even without noticing or hearing about it (actually the main about Gestalt is that all its principles are already known by anyone on a intuitive level).
Now we can summarize the entire discussion into one simple question. Which type of information grouping should the color code privilege? In the end, it all boils down to a cultural aspect. I don’t know which countries are you guys from, but I noticed that at least in USA the people have a tendency to value upsets, at least for publicity reasons more. In my country, however, this doesn’t happen too much. So, I am biased to say that, as pretty much anyone else will be, but to me GreenAsJade way of grouping information is more intuitive.
Your picture is a perfect representation of this subject, thank you.
You misunderstood what I was saying (a recurring theme in this thread). Let me rephrase the sentence:
“My argument is that upsetting/expected game results USED to be a function of the bar chart, whereas currently, they [upsetting/expected results] are not signified by anything.”
They used to be represented by the shades of colour, now they’re not represented by anything.
Fair enough. Let me just restate my assertion that GAJ’s way of grouping would still be preserved if the colours were switched back (as in the left example in your picture), by way of right (stronger) / left (weaker), as explained by the caption underneath. The chart would gain meaning without losing any. Currently (example on the right), upsetting/expected results are not represented by the chart.
Without the bracket text, you do have to consider that this sentence is missing enough information to allow anyone to reach that specific interpretation. And even with that, see below.
No, it would not. In this situation the color contrast is a more defining aspect for Gestalt, so with your suggestion, GaJ’s way of grouping would not be expected. In fact, it’s the same way that yours way of grouping is currently facing right now.
For clarity/correctness: there was no “Old” pie chart.
The pie chart is a new feature.
There was an old way the colours were mapped, which was used in the bar chart.
The advent of the pie chart remapped the colours to achieve best visualisation of the important ratios.
( to a more sensible mapping 
<<< I never said that !)
GaJ
1 Like
Oh, really I didn’t knew that. From what I understood, it looked like soon after you launched the pie chart you changed the color code being used. Thanks for correcting me. Anyway, it could also be considered as the old color code, even when this interpretation is a bit misleading.
From a purely aesthetic perspective, I prefer the current layout, as having the saturated and unsaturated colours opposite each other rather than next to each other looks very jarring to me.
1 Like
To be honest I agree with Bhydden, as in that’s the reason I deem most important
1 Like