Yes, but going way back in time, my theory was that it was based on the number of signups (in this instance there are 53 signups in pen E).
It didn’t deserve to live lol
1 Like
Now I’m really confused…
Score changes from yesterday’s two events haven’t been tweaked overnight.
Despite the TT giving me +5 to 192 and the Zmonthly giving me +1 to 188, my Racing Score today is 188.
I was expecting ~193.
Even though 22nd/55 in the Zmonthly is a better position finish by % and surely should score more than 6th/11 from the earlier TT
This week I got a score increase of +2 for a 30th place out of 50.
So it’s. Not just top 50% anymore.
The scoring is completely broken…
Just won a race 1/118 - 6 point gain
2/15 last week - 17 point gain.
ZRS is going to be like triggers broom (niche UK reference), the seeding mechanism has completely changed, the pen splits have changed and the scoring will need to.
It’s almost like those deploying this have no idea what they are doing and how the product is used day by day by actual customers. The actual long term components come from the community feedback.
At some point zwift have to engage with the community rather than hiding behind ZI articles. Think of the time saved on wasted development over the last 2-3 years… my mind blows…
Happy new year 
3 Likes
This was the Fresh Outta Stage5 Coastal Crown race. There were 63 signups in D. 30th place would be just above the halfway point.
That’s because nobody at ZHQ is a serious user of their own product.
1 Like
Same here. 1/107 and gained 8 points.
Read “Ship of Theseus” for non-UK peeps and/or folks with a classical background. 
I’m beginning to wonder if the Trueskill (or variant of) ranking system being used simply doesn’t work for multiplayer racing games like Zwift, but works fine for say a multiplayer first person shooter like Unreal Tournament.
This is presuming that any ranking system coding written by ZwiftHQ is written correctly.
In a free for all 4 player instagib deathmatch in a suitably sized map, under normal circumstances, the massively stronger player (A) is more likely to come across and kill players B/C/D (2nd-4th strongest). I’d expect A to reach the frag target for the individual map match quickest, or get the most frags in a time limited match.
So at the end of the map match, I wouldn’t be surprised for A to get a big rank score increase in such a small field.
But in a larger field, under normal circumstances, B is more likely to come across the weaker players rather than A and so accumilate more frags and possibly even win the map match. Or A might reach the map frag total target without ever fighting against B. In such circumstances, I could understand A or B getting a smaller rank score increase compared to winning that original example of against three other players including A.
But this logic doesn’t work for a bike race when the only scoring is relative position crossing the finish line.
In a bigger field, ranking score increases ignoring 5min power improvements, should give larger rank score increases to the top approx 50-60%.
Races with less than approx 5 racers finishing should not give rank score changes at all.
1 Like
So Eric Min and James Bailey and no one else? The CEO has no involvement in day to day changes (yes I’ve chatted with him in-game) and James is great at event stuff but doubt he has much or any involvement with PD or ZRS. I know you’re joking about calling the guy who created Zwift being at level 37 qualifying as a “serious user.”
I’ll continue to say “no one” at ZHQ is a serious user because two is close enough to zero, and those two have zero to do with development.
1 Like
My question to Zwift after they proudly said they hired Phd Scientists for ZRS was ‘Nice, but do they cycle, do they use Zwift and do they race’.
Guess we know the answer to that now.
1 Like
test environments exist, Jon likely doesn’t make his rides public. Multiple people at ZHQ ride in the test environment and can’t share the rides/make them public.
1 Like
That’s a stretch.
Until anyone at ZHQ demonstrates deep understanding of how literally anything works with Zwift racing or dynamics I will continue to say that no one at ZHQ uses Zwift seriously. If you want to be mad about it that’s your choice.
How does “serious users only in a test environment though and that makes them qualified and informed on racing” work out? How do these test environment races work?
How does the complete lack of QA mesh with this test environment seriousness?
2 Likes
That’s not using like a paying customer…
It’s like the CEO of a company always thinks all of their buildings are freshly painted - what they see and interact with is usually a far cry from the coal face.
Let’s just be honest about things, the race community is a little bit niche and not necessarily atypical social user of the platform… Dropping into a race once or twice probably doesn’t show you the issues of those who race daily or weekly see.
If they understood how their decision will impact users we wouldn’t have got ZRS v2 as that was a disaster that showed how out of touch they are with niche users. ( I say niche, women and light weight riders really aren’t niche but they are a small proportion of racers)
3 Likes
Remember how they fought for months to make the 10m/30s power numbers work despite being told by their smartest and best-looking users that they were using irreparably broken metrics and needed to switch to using a five-minute compound score instead?
And then eventually threw out everything they paid the PhD data scientists for and came around to using the 5m compound score in the end?
Good times.
It might be as simple larger fields give greater uncertainty of result so less score movement is applied, whereas small fields provide a greater certainty of result so larger movement of score is applied.
Now to me that doesn’t really work out in the field, but that is a guess as to why it works the way it does.
What we need is some transparacy and probably an indication of how zwift think it should work - that’s the key bit, as if zwift and their users are going in opposite direction with how they think it should work then it’s going to be a roller coaster for a while.
1 Like
It’s going to take a while but I’m confident that ZHQ will eventually abandon open skill and go to an ELO system as their best and brightest users have encouraged from the start.
We need the popular race series like the Zmonthly and Tiny Races to regularly include hills that will take all non-alien levels at least 5mins, such as…
Petit KOM,
part of Ven Top (~3Km for pen E, but progressively longer to ~6Km for pen A),
part of Innsbruck KOM,
part of Epic/reverse KOM
etc.
1 Like