My crack at a Mitch Hawker style poll

[JJLehto]

Hey CF, I am not sure how big a deal the Mitch Hawker poll is here, but as you all surely know it has not been done since 2013. Sad, especially since so many RMCs and intriguing coasters have opened since. I have looked into why Mitch has stopped and seems no one knows, nor did he respond to an email I sent, (he did once before so he may truly be done). Anyway, I have a break from school in 2 weeks, so I am going to take a stab at continuing the poll, with all credit given to Mitch of course.

I was thinking of using the same technique he does, as listed on his site, perhaps with tiny tweaks here and there. I am here to ask:
1: Would anyone participate if I did this?
2: Does anyone have any ideas, small or big? I am willing to listen to all ideas.
I am gunna ask a few enthusiast sites and see what reaction I can get. I loved the MH poll as did many others, I would love to see it continued and am willing to give it a try.

 

[Sythe]

I would participate. Don't really have any ideas though.

Sent from my Moto G (4) using Tapatalk

 

[Pink Cadillac]

It needs to be quicker and less faff for participators.

 

[Hixee]

^Exactly.

Mitch Hawker became such a burden once you had over like, 50 creds. Once you're into the 250+ category, it's just awful. There's no way I'd be able to find the time, or the motivation, to sort them all out. I suspect that's why it died out.

To be honest, @valleyfair!'s "Annual CF Coaster Tournament" (based off a bracket system) have proven to be rather good over the last 10 years. You can see the summary topic and final round here, but a quick search of the forums for his username should pull up the other rounds:
http://coasterforce.com/forums/threads/2016-coaster-tournament-finals.40421/page-2#post-971822
http://coasterforce.com/forums/threads/10th-annual-cf-coaster-tournament.40313/#post-968551

 

[TMCoasters]

I would definitely be willing to participate. Good luck with it!

 

[gavin]

I didn't bother voting in it the last time, and I've got way more coasters since. Honestly, it's just too time consuming. Plus it becomes incredibly frustrating to see a lot of stuff I've ridden and voted for unlisted due to lack of riders. I understand that, obviously, but it doesn't motivate me to bother knowing that a lot of my placings won't count for s**t.

 

[JJLehto]

That's fair, whatever I try to come up with will likely include some scaling, maybe, to take riders into account but not exclude any. Mitch also seems to take rider experience into account, so I'd likely do the same. Basically if you're one of 4 who rank a coaster it'll be on there still!
And the time consumption is very reasonable. Took me a bit and I was sub 100, with 200, 500, 1000 I can't even imagine and frankly I'd only do like a top 50 and leave the rest. I think I did that myself, just kinda half heartedly shuffled a lot of the middle ones around.
It's something I'd also take into consideration, either cutting down the #, finding a way to make it easy or just stress no pressure to rank all 1200 coasters which would be a truly mad task

Anyway, I just am gauging interest on some forums. When I come up with a test run, if anyone would be willing to participate it'd be helpful

 

[balrog]

Wouldn't it be possible to make a ranking based on random one on one match-ups and a chess-like ranking system ?
I guess an easy way would be to team up with coaster count to generate match-ups within the list of coasters one has riden.

It is much more efficient than braket style tournaments and MH style polls, while giving rather accurate results.

 

[JJLehto]

Coaster Count is something I've thought about it to make it easier, and a head to head like that could work, seems like what CF does in a way

 

[balrog]

The difference is, you cannot get a ranking out of a braket style tournament. ELO-style rating allows for a total order, and using a relatively small amount of individual match-ups

 

[Pink Cadillac]

You could keep it keep it the way MH was but limit the number of coasters that are ranked. Using the coaster tournament we have here for example, the amount of coasters in the poll is 128. The list is not extensive but covers almost everything relevant and you might be able to do something similar. Most people don't care about MH's list outside the top 50 so there isn't much point people ranking every family and kiddie coaster they've been on. Even a list of 200 will significantly reduce effort as most people would have ridden less than half of them. Hope I made sense!

 

[balrog]

I'm going to look insistent, but with random match ups, you could have a complete ranking while only asking people a dozen of simple one vs. one questions. I really don't see the point of keeping the MH style poll.

 

[Hyde]

The advantage to the MH-style poll was in being a cross-wise comparison, it allows for "apples to apples" comparison of roller coasters across those who have ridden them.

Anyone who has downloaded the official results will notice a large excel sheet showing wins and losses for each roller coaster, based on if riders ranked a roller coaster above or below the correlated roller coaster. This is to say, the spreadsheet takes Nemesis, and compares it to all the other roller coasters on the list. For those ballots that marked Nemesis above the respective roller coaster, Nemesis gets a win. Those where Nemesis was below, gets a loss.

The downside to cross-wise comparison is larger ballots get greater favoritism to larger ballots. That is, if you have ridden more roller coasters, you carry greater sway in how each roller coaster performs in the comparison. MH tried to mitigate this by cutting off rides with low ridership *coughTPRAsianCoasterTripscough*, that helps eliminate selective bias. At the end of the day however, larger ballots did have greater choice power than smaller ballots.

That is where the notion of capping at 50 sounds like a good idea. It means each ballot is the same length, and also helps create a greater sample size for each roller coaster; the more ridership per ride, the less likely bias and variability is to occur.

I have thought about trying to remanufacture the MH polls to do as an official CF effort from time to time; however I have always enjoyed the simple romance of Valleyfair's straw poll too.

At any rate JJLehto, I'm happy to help out. Message me and we can talk more.

 

[gavin]

The downside to cross-wise comparison is larger ballots get greater favoritism to larger ballots. That is, if you have ridden more roller coasters, you carry greater sway in how each roller coaster performs in the comparison.

That was never a downside. That's what made them the most accurate polls we had.

People just get pissy when they see rides they haven't ridden do well in polls.

Without that system, you end up with s**t like the Golden Tickets where only number of votes counts.

If you've got 1,000 people who've ridden Kinga Ka and 100 who've ridden Kinda Ka and Dodonpa, most of the 100 will say Dodonpa is better, yet Kinga Ka would win. It's stupid.

Every year with Mitch Hawker you'd get a bunch of piss babies crying it's rigged because "some Asian coaster with 20 riders is higher than my favourite at Cedar Point". Yeah, because it actually Is better and those 20 riders will have done both to know.

 

[JJLehto]

Thanks for the input and while the bracket idea is intriguing I have to agree, MH did it best and my intent is to basically continue it. Hyde and Gavin already explained it wonderfully so no need to rehash. And ill send some PMs soon. On the tech side I may need some help and I need "test subjects" to make sure results come out OK. The actual data collection and all I think will be tedious but not too awful, I already do stats/economics daily so I'm ready ha Expect PMs/follow up in 2 weeks. I have a break from school then

 

[CanobieFan]

Every year with Mitch Hawker you'd get a bunch of piss babies crying it's rigged because "some Asian coaster with 20 riders is higher than my favourite at Cedar Point". Yeah, because it actually Is better and those 20 riders will have done both to know.

Yeah but you had TPR trying to sway the vote in order to sell spots on trips....(hey look! All these rides hit the top 10 of the MH poll! AND WE WANNA BRING YOU THERE NEXT SUMMER!!!! Now hand over $1,700***) ...And you also had OnRide.de rate bombing TF out of Superman New England to get GeForce back to #1..... Suddenly ballots were coming in with only 8 or 9 coasters rated but not at consistent parks and suddenly 'my favorite SLC' was placed number one and Superman was placed number six. And the other coasters would just be random handpicks.......

Maybe if we tag @rollermonkey enough he'll return and help shed some light on the demise of the poll. I believe he was actually working internally with Mitch.

*** As someone who was part of the inaugural TPR trip. And felt the personal backlash/fury for not enjoying Balder better than life itself.... But that's a story played out some 11 years ago.

 

[Hyde]

^ [EDIT] - @CanobieFan Beat me to it!

That was never a downside. That's what made them the most accurate polls we had.

People just get pissy when they see rides they haven't ridden do well in polls.

Without that system, you end up with s**t like the Golden Tickets where only number of votes counts.

If you've got 1,000 people who've ridden Kinga Ka and 100 who've ridden Kinda Ka and Dodonpa, most of the 100 will say Dodonpa is better, yet Kinga Ka would win. It's stupid.

Every year with Mitch Hawker you'd get a bunch of piss babies crying it's rigged because "some Asian coaster with 20 riders is higher than my favourite at Cedar Point". Yeah, because it actually Is better and those 20 riders will have done both to know.

What I was alluding to was collusion of the TPR asian trips. While yes, some of those Asian roller coasters are indeed amazing - seeing Kawasemi clock in ahead of Nemesis with 19 riders did not feel like an appropriate inference. If Alvey were less forward with encouraging his followership to vote his way, it'd be more palatable. But politics certainly drew question on sample bias.

 

[CanobieFan]

But politics certainly drew question on sample bias.

And at the end of the day, it always came down to politics. Golden Tickets are no different. Same really for any other poll. But I've always felt I don't need someone else to tell me what ride to like. I'm fine and able to decide that on my own. Just because you don't like a certain ride doesn't mean I can't like it. And just because you don't love my favorite coaster doesn't not make it my favorite.

 

[Hyde]

Yeah, the Golden Ticket Methodology has always been baffling. They use a total point system of coaster enthusiasts throughout the nation that simply are not well travelled. (Hence the anti-Euro bias) Total Point System also means those that win, win big (1st place gets 10 points, 2nd gets 9 points, etc.). Whereas the MH pairwise regression (I called it cross-wise earlier by mistake) allows for greater parity of roller coasters. Millennium Force doesn't win big in Golden Tickets because it is ranked highest in voter's ballots; it wins because it is ranked most often, and always accrues points.

 

[davidm]

Was it that blatant (Ok I see Canobie's onride.de point and that does sound blatant) ; isn't there just some in-built coaster nerdism that says if I've ridden a ride that you haven't then of course it was brilliant and you just don't know because your such a noob?

The MH poll must have been hellish to compile though because of all that "mutual rider comparison" that drove it - some very complex back end stats churning going on there I always felt. I did think it gave a pretty good view though - perhaps there should be a some way of objectively excluding obviously weighted ballots (perhaps there was?). Goons will always try to game the systems tho' so perhaps its a forlorn hope.

Anyway as Absolutely Reliable always says ; 'your favorite coaster sucks' and I also don't need no-one to tell me what to like!