Honestly I've had this feeling as soon as I started playing this game, but I feel like the in game timer is off -- and I will prove it.
After watching F1s run I don't see any reason to believe that that boost in the 2nd lap's tunnel was good enough to save 1.7s over me, even after many mistakes during the run. He slows down at the quick U-turn twice, which is at least a second per lap. His ending section is as tragic as my old run, if not worse. He goes wide many times and has bad bridge turns. So. Would that lap 2 tunnel - wide ending - bumps - bad u-turns - bad bridges = a 1.7s timesave? I very much doubt it.
But if that wasn't enough: Here's a true doubter for you guys. I've commentated a comparison video comparing my (current) PB, this next video, and F1s vid, but I cba rendering that and uploading it at 5 AM.
3:28.8 igt run.
Looks like shit got a lucky ass boost at the end but
This was played in very shitty laggy conditions. Then sped up in a video editor to have it be close to the original speed. As you can see, the cutscenes are ridiculously fast. I even crash in the last section. Yet it's still 6 seconds faster. If anyone truly dares to tell me that the rate of playing the game doesn't matter now, you'd be clueless. IGT is skewed. I'll post the comparison vid tomorrow.
How do you propose we fix the timing for runs if you feel IGT isn't fair? IGT is good because it doesn't matter if someone's flash player lags a little bit. Although -- I don't know how I feel about playing the game at a slower rate intentionally to abuse IGT.
This makes me sad mate :( You seem to be correct. What did you use to generate the lag? Pour one out.
So... is there a reasonable solution other than RTA? Seems unlikely, unless goomba's generated lag is somehow different from "normal" lag. If RTA is the only option then goomba has wr by like 10 seconds, and I could hop on a friend's battlestation desktop and bop everyone else in a few minutes. Both IGT and RTA appear to not reflect true skill.
Is there a third option? Some way to count in-game frames (and would that even help anything...)? How much BS is worth sifting through to make this game fair to speedrun. We can't exactly TAS it or easily log our inputs on a frame by frame basis.
How bout we get in touch with the officine pixel devs and beg them for the raw .dir files from 2004. Either they somehow have them or the game dies. It's a win-win!
First of all, thank you for this Goomba. I wachted the video of F1 as well and decided to compare it to my run. Guess what? My run was probably better than his! I started both the video's when the light turned green and I entered the train tracks at the end around 3 seconds earlier than F1. Then how come my IGT was almost 2.5 seconds slower than F1's? I think the IGT is broken as well.
Before we get ahead of ourselves, I do want to point out why IGT is used in the first place. RTA times are wildly different between runners depending on their setups. For instance, my run is about 12(!!!) seconds faster than f1s, 10 seconds faster than yours, and about 8 seconds slower than goomba's, putting an astronomical 20 second gap between f1 and goomba. Let's not kid ourselves that RTA timing is more fair given our current pc setups.
@Wixirix the game runs at different speeds based on hardware, but that's not the issue here. I still think it's slightly iffy that you can run the game at such a slow speed compared to my normal state, and still be able to compete. Nobody would allow slowdown in super mario bros (if it was run by IGT), for example. Imagine a fighting game where 1 player can fight another person at 80% speed. That'd be ridiculously broken lol but technically this is not the biggest fault.
The error lies in that the timing doesn't seem to be accurate with slowdown -- I really don't know which way the cookie crumbles. What I do seem to find is that a faster PC doesn't seem to have an advantage in IGT: Most runs that I've concluded as -should've been slower- were "slow" RTA times, and that's not just because the game runs slower.
I've noticed that pushes at low speed seem to be a major key. We could compare a "normal RTA" pushless run to a "laggy RTA" pushless run, and see if the difference lies within the cutscenes or not.
Made a discord so we can discuss things there
Can we keep the leaderboards as they are for IGT sorting, add a second column for RTA times, then an "Accuracy" variable stating whether or not the run should actually be what it was? And the only option for that should be Inaccurate. I'm thinking something like this; 
