Wheel diameter, mirror shape, and two rear license plate holders were critiques I didn’t expect. Seems the Moxie’s only strength was its drivability.
Reviews: Part 3
Another day, another stack of papers and another tight section of the ranking.
#13 - TAE - ECC 144 @breadtheloaf
A poorly and hastily styled entry, with no effort at detailing or surfacing and discordant graphical elements, especially at the rear, somewhat salvaged by a decent front end and a year-appropriate body. Engineering-wise, it’s a mixed bag, with decent tuning but confusing component choices, like adaptive dampers and active anti-roll bars but recirculating ball steering and solid disc brakes, all of which would have constituted an elimination if there had been any. Its stats are propped up by its decent safety and great environmental resistance, though reliability is near the bottom and other stats are generally below average.
#12 - Waldersee Attache 1.6i @Texaslav & @Maxbombe
One of two perfect design scores in the competition, high on effort and creativity, with realistic surfacing and detailing, and complex graphical elements, especially the rear end, with its two-tier wraparound glass. With all that, its sad to see it so far down, as its engineering statistics aren’t the greatest, even though its component choices are decent and realistic and tuning is well done. Its drivability is below average, as is its comfort, with a focus on sporty dynamics worthy of its design. Reliability is great and costs are good, though practicality is on the lower end and, overall, in such a tight pack, its score just can’t get much higher.
#11 - Marnel Verno Life @superbiirdd
A cutely-styled, brightly-coloured, egg-shaped little car, with great detaling and surfacing up front and a decent execution everywhere else, making this one of the most interesting looking cars in the competition. Although it is decently engineered, with realistic component choices and performance, its suspension tuning lets it down, as its stiff rear springs make for a queasy ride, resulting in poor drivability and below average comfort ratings, as well as the second-lowest sportiness rating. Practicality also comes in at the lower end of things, though it’s not terrible. Its reliability and safety are decent if below average, but its low purchase and service costs somewhat make up for its shortcomings.
#10 - Alvern Bravo SE @Capri78
A surprisingly complex design that gets lost in a sea of black, with decent shaping, surfacing and detailing and no glaring omissions. On the mechanical end, its poor drivability disappoints, but it’s decent everywhere else, with good reliability and safety. Comfort is above average and its costs, as well as its sportiness rating, are right in the middle of the pack. The engine is well tuned, albeit with some needlessly strengthened internals and highly complex emissions equipment. Elsewhere, component choices make sense, though tuning-wise, the rear suspension is extremely stiff when compared to the front, which no doubt hinders its driving dynamics.
8 Likes
10th place, good stuff. Didn’t expect I’d end up that high.
Also the reason it’s so black, is to try and hide all the imperfections in the design, mainly with the 3D Roof.
3 Likes
Reviews: Part 4
#9 - Harris Sunday @David_Herrera
This design is good, if not amazing, with era-appropriate moulding and surfacing and a nice looking front end. The rear is also decent, though the high-mounted taillights could have been worked on a little bit more, as they are somewhat too far inside the body. Generally, it looks good but it’s not groundbreaking in any way. The same can be said for the engineering, where it only excels in its safety rating. Its comfort, practicality and sportiness are above average, but its environmental resistance is the lowest in the competition, and its reliability is also on the lower end of things, likely due to some of the more advanced component choices. Even with its double wishbone front end and its limited-slip differential, its drivability is quite poor, and the cost to purchase and to service it is on the upper end.
#8 - Kamoto Millenia Final @CrazyCat808 & @NotChris07
The second perfect design score in the competition, with incredible surfacing and detailing as well as attention to detail in the graphical elements, like the front and rear lightbars and the lower bumpers. Its engineering, however, doesn’t quite live up to the standard set by its styling. Although its components make sense for the segment and quality is well-applied, its poor suspension tuning, oversprung at the front and overdamped at the rear, makes for an unstable ride, resulting in poor drivability and below average comfort. Purchase cost is on the expensive side of things and its practicality is at the lower end, but its reliability and safety are near the top. Overall, it’s kind of a mixed bag, which despite its excellent styling doesn’t manage to climb as high in the rankings as it could.
#7 - Norrsken Felid (Exploro 1.6E) Clone @Ritz
A nicely-detailed, realistic and era-appropriate design, though on the simpler end when it comes to surfacing. Overall, it’s easy on the eyes and perfectly appropriate for the segment, even if it’s not a radical proposal. Mechanically, it sports an overbuilt and decently powerful engine, more powerful than other entries in this size category. Its component choices are appropriate and realistic, and its tuning is generally good, except for the overly stiff rear dampers. Its drivability and reliability are stellar, but its purchase and servicing costs are right about average. Elsewhere, it scores decently well, if not excellently.
#6 - SYEON Pokke 1.1 Simple (UKDM) @supersaturn77
A cute, small city car with excellent detailing and surfacing up front, with a charming, almost smiley face look. Though the sides and rear are weaker, the overall look is still one of the best in the competition. Where the engineering is concerned, it employs a solid mix of high-tech, high-quality components, which makes it somewhat expensive to purchase and service for its size and performance. Great tuning and calibration results in decent stats overall, with its strong points being drivability – the best in the competition – and reliability. Its only weak points lie in sportiness, which is just below average, and comfort, which is average.
#5 - Oni Nodachi Type E @Riley
Definitely a good-looking design, with a detailed and appropriate front end and good surfacing, moulding and detailing. It’s not without fault, however, with a lower rear bumper that doesn’t quite match the quality of the rest of the design, as well as extremely wide fenders that show its high-performance origins but don’t quite fit the economy car look. Falling on the compact segment, rather than subcompact, this is a larger and more expensive car than the rest of the competition, using up the entirety of the allotted purchase cost budget. It also goes against the flow with its longitudinally-mounted boxer engine, which makes less power than one might think for an engine of its size and the size of the car it’s installed in. Its only weak points are comfort and practicality, due to its stiff suspension and bodystyle, but it scores well, if not excellently, elsewhere, with its strongest point being its safety rating.
9 Likes
Reviews: Final Part
The time has come for the team to choose a winner, leaving the four best scorers for last.
#4 - Alouette Primo @mart1n2005
This tiny, quirky and sporty hatchback is definitely a looker, with great attention to detail when it comes to surfacing and graphical elements, without any glaring weak spots, though it might be a bit on the sporty side for what the segment is about. Some of the mechanical components chosen are on the higher end, namely its variable electric power steering and electronic stability control, which results in a somewhat high cost for something this size. Its engineering focuses on good dynamics, with good drivability and decent sportiness, without sacrificing comfort, which falls right in the middle of the scoreboard. Elsewhere, it has great practicality and its reliability is good but average.
#3 - Breda Lumo 60 @VanZandt_Breda
Another tiny egg-shaped hatchback, with short overhangs and a tall, spacious greenhouse. Detailing and moulding is simple on the sides, but the front and rear ends are well-finished and creatively styled, especially the rear, with its two-tier window and full-width taillights. Engineering-wise, it achieves its stats mostly through quality rather than the most high-tech components, resulting in incredible reliability. For the same reason, the cost of entry is near the limit, though its servicing costs are reasonable. Drivability is just above average, while safety is good but average. Elsewhere, it scores well but not incredibly so.
#2 - Nordwagen Skadi D 4x2 @moroza
A larger, more refined-looking proposal than most other entries, though it does look a bit older than the model year suggests, even without the heavy throwback elements. Still, it’s all executed well and it shows creativity and attention to detail. Mechanically, it is the second longitudinally-mounted boxer entry. It combines a realistic mix of components, a careful application of quality – both positive and negative – and good tuning, to great results. It manages to score well everywhere even if it doesn’t excel anywhere, all at a reasonable price point where more expensive entries don’t manage to score that well. It does have weak spots, however, namely in comfort, sportiness and servicing costs.
#1 - Uanoi Bonkly B ThermalFlux v1 @LennoxV10
A terribly named, hastily styled entry, with very basic fixtures and not much in the way of detailing or surfacing. Mechanically, it wins by score only, leaping ahead of the rest of the pack in safety and comfort, where it is at the top of the list, due to its soft suspension and premium interior components. Reliability, practicality and environmental resistance are all good, while purchase costs and drivability are decently good. Finally, its sportiness is the lowest here, owing to the soft suspension tuning that gives it its comfort. Tuning of its small turbocharged engine is poor, with a mountain of boost at the middle of the rev range. Due to the use of turbocharging, its service costs are the highest in the competition. Although unscored, it achieves incredible fuel economy. All in all, its top-tier scores in mid-level priorities, combined with its decent scores in high and low-level areas give it its winning score.
Scoring Spreadsheet:
7 Likes
Curious about the scoring? Yeah, I was too, and the results here are the result of an… interesting design decision.
The spreadsheet hasn’t been changed the normal way, as an interactible spreadsheet, but rather using what I’ll call a “pubhtml” URL. You can change the end of that though, so we swap to xlsx and we get https://docs.google.com/spreadsheets/d/e/2PACX-1vRpDlpXmh-FIQgULaKVGdNuC9QI3uphktuT-HAEO5T4LKM0zQXjctoHPMBOaFcHYg/pub?output=xlsx
This gives us all the formulae which were used. Chowi made a deliberate decision to have entries receive points based on dividing the entry’s score by the best score around. Best in a stat gets full points, while you have to drop to zero to get zero points. Stuff where lower is better is scored by the reciprocal of the stat.
The problem with that is that it fundamentally de-emphasises stats with low field spreads relative to the maximum. The maths is fundamentally broken for delivering on the brief. By being an outlier in low-star priorities, the winning entry has fundamentally increased the value of those priorities. The second worst entry for drivability still has 75% of the best, and the second worst entry for reliability still has 86%, so these stats are really worth 0.025 and 0.014 rather than the expected 0.1. Meanwhile, with a massive outlier for comfort, you wind up with the second highest having 74% of the best and the average having 61% of it, massively inflating that to be worth something like 0.02 if we run by how many points it gained over second-worst.
That’s right, in terms of gap between first and second worst, comfort was more important than reliability here! Because the scoring system just didn’t consider how field spread may differ between different stats and how it’d fare against real cars!
(And in terms of gaps between top cars, it’s even worse, but that’s a consequence of linear scoring without properly handling outliers)
7 Likes
yeah I was really wondering how that happened
2 Likes
Super interesting!
…and it turns out even that isn’t the end of the errors. Because there is an error in the formula that means that, instead of the 0-1 comfort score being multiplied by the weighting, it is added to the weighting, because there is a + in the formula at that point where there should be a * !
How much difference does this inadvertent error make? This adjustment alone bumps the “winner” down to 18th, which is clearly far more in keeping with its below-average scoring in the categories responsible for 65% of the weighting. With this fix made, the actual winner is the @Texaslav x @Maxbombe entry that landed in 12th in the erroneous standings!
(I was checking these standings vs. how the same entries would stack up with the scoring I’ve planned for QFC70. The r2 correlation coefficient between how I ranked them this way and how they’ve been announced here is less than 0.3; with the fix applied, this rockets to around 0.9. TLDR; the two sets of ranks barely agree with each other if unfixed, but with the fix they actually do.)
Here is the scoring spreadsheet with and without the error: TMCC40 scoring and it's the same but there's no formula errors so it's not - Google Sheets
11 Likes
uh wot
2 Likes
Interestingly, my position based on this, is where I actually expected my car to be, 10th did feel oddly high for an entry that was lower-midfield at best.
What I find interesting aswell, is with the error, the sheer gap between P1, and P2 is basically 10 whole points, the largest by a mile, but with the fix, the gap between P1, and P2 is just under 3.
The other thing that does confuse me, is given that design/execution is a joint highest priority, how that car ended up being 1st, by a massive margin and not seeming suspicious, whilst having a design that, whilst not entirely terrible, was lacking in many ways.
1 Like
Confirmed, just found the same thing goofing around with different scoring approaches. It’s “O25+$N$1” when it should be “O25*$N$1”
2 Likes
Damn I goofed hard
I apologise for the mistake, that’s what I get for rushing the spreadsheet and not checking formulas. Flavor text will remain mostly the same, but rankings will be updated soon.
2 Likes
Reviews: Oopsie Daisy
After the team looked over the results, someone noticed that one of the interns had made a mistake when calculating the scores, which completely changed the rankings, in some cases by a significant degree. After looking it over and fixing the mistake, these are the resulting rankings:
- @Texaslav @Maxbombe
- @moroza
- @CrazyCat808 @NotChris07
- @supersaturn77
- @superbiirdd
- @mart1n2005
- @Ritz
- @Marv666
- @Isabella
- @Riley
- @VanZandt_Breda
- @AKA_NOBUDDY @Jaeger
- @Kreator
- @feffel
- @David_Herrera
- @Capri78
- @Ultimate_Billy
- @LennoxV10
- @Happyhungryhippo
- @abg7
- @ArizonaCaseo
- @AndiD
- @breadtheloaf
- @PoseidonAutomotive
Huge apologies for the mistake, especially to anyone that got excited or disappointed, will do better next time. I just rushed the spreadsheet from something I had done for a class I never cared much about, carrying the mistake all the way to here, and never checked it over afterwards. As for changing the statistic relevance of stats, that’s definitely something I’ll try to look at for further challenges but not something that will be changed for this round.
Scoring Spreadsheet:
10 Likes
oh my god
AND THAT’S THE BOTTOM LINE!
3 Likes
Hopefully this serves to demonstrate how much easier it is to have a good idea than to implement it with all the nuance a good idea deserves. I know I am indebted to a lot of smart people who know how to build a clever spreadsheet that handles all the math that is a bit over my head.
I’ve seen a few people here and there who are happy to share their raw formulae and well-designed spreadsheets. It may be worthwhile for some of the heavyhitters of the userbase here to spread some of that knowledge around and maybe in time we can develop a collaborative culture around building these spreadsheets that really enable fun and satisfying contest parameters.
Thank you, the-chowi both for hosting and being open to correction in this regard. I hope you don’t feel too bad about this mistake, as I think a lot of us would probably find a way to mess up a detail like this also. And good work, AMuteCrypt and ShinyBat for putting together some handy descriptions and examples of how it went wrong.
8 Likes
Even though I’ve only hosted 4 challenges here, I’d be more than willing to contribute to the spreadsheet cause. Maybe in the future there will be a standardized, customizable template that hosts can use and modify to their liking.
3 Likes
Shit happens. I still feel bad for the time I mixed up two stats which made the winner (@moroza ) not even go further to the finals at first, which fortunately enough was pointed out in time. But humans will make mistakes and after all, winning here is only about honour and confirmation that you understood the brief well and made a good attempt to adapt to it.
5 Likes