It has been brought to the mod team's attention that a recent run submitted to this board has been alleged to be a TAS, a tool-assisted superplay, passed off as a legitimately human-performed RTA run. The evidence that has been presented by the accuser will be shown here for your review. They are free to put their name on their comments in this thread if they wish. Additional evidence has been gathered upon further investigation of the matter. My editorial responses to these points will be presented in [brackets].
The run being questioned is the current Endless, 99,999 points Hard record by Raidabel0w: https://www.speedrun.com/panel_de_pon/runs/y86gk9nm.
Point #1: This is the account's only submitted speedrun to this website. [It is not the user's only submitted speedrun to this website. Previous runs to this leaderboard have been submitted under the name Rika_Furude, an account that has since been deleted--see the first place 99,999 points Easy run: https://www.speedrun.com/panel_de_pon/runs/ylne2rxz. Generally, a world record run should be under a greater degree of scrutiny than most other runs. They reflect on the community as a whole. I did not verify this run so I am not aware of the amount of scrutiny that it went through during verification.]
[Here, the names Prierè Tsumerai Ame, Rika_Furude, and Shippai all intersect with each other. Remember these names for later points.]
#2: Timestamp at 0:10. The HI-SCORE is 2000. Prior runs from this user have shown that they soft reset, showing a high score of 99999. See this run: https://www.speedrun.com/panel_de_pon/runs/mkv2o21yhttps://www.youtube.com/watch?v=88nvYi4z3Uk?t=9
They either were hard resetting, or this was their first attempt from boot. [You must game over after power on to change the HI-SCORE. If you pause and quit, your HI-SCORE does not update. If you reset before game over, your HI-SCORE does not update. One could quit the run if they know that it will not beat their best time before it finishes. Resetting between attempts would be tedious, but not unheard of. I often do this for a better presenting final video, but not always with Endless.]
Due to SRC thread limits, the content will continue in the next post.
#3: Timestamp at 0:10. The user is messing around with DAS [delayed auto-shift] during the countdown instead of positioning their cursor in an advantageous position. [As someone who has traditionally relied on DAS, even I will admit that DAS is not particularly useful for side-to-side movement, but can be for the up-and-down movement that is being mostly demonstrated here.]
#4: Timestamp at 0:13. The user does not hold down the raise button, but instead presses the button for one frame every time the stack raises one layer. This occurs every instance the stack is raised, and sometimes even alternates between the left and right bumper. Additionally, the cursor remains in place in the corner instead of positioning itself to where it can access more of the stack (such as the center) ahead of time while the stack raises.
[This is built on more with the next point, but I will agree that this is not the typical way to play. Most runners I have been around--and I have been speedrunning this series of games since 2013 and seen a lot of people come through this community--will hold one bumper to raise the stack and place their cursor in a spot that they want to start swapping when they finish raising. Another point I will add here is that most top players will generally raise their stack while flattening it to a level state because you need to get the maximum panels on screen to maximize your chain as fast as possible. This user does not ever do this for the duration of the run. For an example of the typical runner experience with Endless, check out our community's most recent Endless speedrun tournament for Tetris Attack, a functionally similar game:]
#5: Timestamp at 0:15. Cursor speed is outrageously fast and smooth throughout the entire video. Throughout the chain, the cursor temporarily pauses at times instead of immediately moving to position itself for the next chain. [It is hard to believe that the cursor speed can be this fast, yet the mistakes made here can co-exist in the same performance. I certainly have known people that have issues controlling their speed.]
#6: Timestamp at 0:31. User deliberately puts in mistakes in what appears to be an attempt to show that the run is not "perfect." [A better example, in my opinion, would be at 1:33 where the user goes on to make a x15 chain despite the x? chain links scoring zero points. To have the 20,000+ attempts as indicated on their timer and the demonstrated cursor speed and either not know that x? chains score zero points or demonstrate that knowledge by stopping prior chains at x13 and continue the chain anyway does not make sense to me.]
#7: These are prolific throughout the entire run. Cursor speed is extremely fast and does not follow normal human runs.
[I was not the one who verified these speedruns. I saw the run in question for the first time last night and physically cringed at how robotic it looked. Community speedrun races were happening at the time in our Discord server and other runners were expressing similar concerns, one going so far as to ask outright if it was not a TAS.]
That is all for the accuser's expressly stated points. Upon further investigation, I was directed to runs by this user actually declared to be TASes:
These are done on the community PvP engine called Panel Attack. It has its own Endless leaderboard here: https://www.speedrun.com/panel_attack
Anyone else see similarities with the RTA runs? Don't look too closely at the cursor speed. EX Mode is an even bigger challenge than Panel de Pon can ever give and demands even more from the player. I see the same clean lack of stack raising while stack flattening. I still see some tapping of the raise button. I am trying to play devil's advocate as much as I can with the allegations, but all of Rika/Raida's runs I can see have the same issues.
I am inclined to remove all of Rika/Raida's speedruns and permanently ban this user from submitting any more runs to our leaderboards due to the prevalence of these claimed RTA runs appearing to be TASes, but this is not a decision that I can come to unilaterally. I am open to hearing why these runs are actually valid RTA runs or why I should not remove all of Rika_Furude's / Raidabel0w's runs from this leaderboard until the end of this month. We will figure out how to proceed if a valid justification for these runs being legitimate exists. If no one comes forward with good enough reasoning, Raidabel0w will be banned from submitting to all leaderboards in the Puzzle League series going forward.
Please post your thoughts in this thread.
I remember someone in the Panel Attack discord named “shippai chi” being knowledgeable with TAS tools. But that was a while ago and I didn’t interact with them personally.
Looking through older videos of theirs to compare - it does appear that they usually raise stack with tap vs hold on runs that are more clearly RTA as well as TAS, so I'm not sure that particular point holds up. That said, the movement does appear to be smoother than standard play. One note is that in reviewing they have some runs listed as "touch control - non TAS" - these reflect a much different play feel than their other presumably controller based runs. I'm not sure if this difference would be enough to account for the unhuman looking "smoothness" of the run in question, but I think it bears review. My opinion based on a watch of all their posted runs and the above points, is that this run MORE closely resembles their previous "TAS" runs than it does their previous "RTA" runs, (based specifically on cursor movement speed, smoothness, style, consistency)
I’m pretty sure it’s the same person. All the points stated above are true. It’s undoubtedly a TAS run. I think a full ban is appropriate.
(Someone else mentioned to me this TAS-looking run to me so I told mods about it) I don't trust my judgement on matters like these, so I really appreciate Cards doing an in-depth analysis.
I can't add anything to said analysis, but something else I found interesting was that 8 months ago (only 2 months before the deleted easy run) , on the same channel "Rikka" uploaded a speedrun with much slower gameplay. Assuming that Rikka Shidai and the channel owner truly all are the same person, if the improvement is legit I'm jealous at the rate of progress and the work ethic required for it.
Also, while some people (including "Rika") say touch controls are straight up superior, I have my doubts on them being better for small intricate movements. IMO, they're a different input style you have to retrain a lot for in order to get results with, they're not an automatic powerup.
I will reiterate some of my thoughts from last night in a more public forum.
Initially when I saw this run, I was slightly skeptical, but not convinced it was a TAS run. However, the more I compared this runner's "RTA" videos with their "TAS" videos, I became more convinced that it was a TAS. The RTA video shows that they do hold the raise button before tapping, so a consistent tap would need more accurate frame-perfect like inputs. Noticing this inconsistency, I then paid attention to the speed of the RTA run vs. the TAS run. Like z2600 said, improving your speed that quickly would either be a war of attrition with your small motor skills, or just not likely. The original comment with the video said he's been grinding this category, but I just don't buy that much improvement that quickly. That point can be thrown out as pure speculation on my part, and that's fine. Another thing that rubbed me the wrong way was his lack of trying to find clears as the stack is rising. Pre-planning your chains is a vital part of speedrunning modes that need quicker reaction times, and the lack of that in the sub-3 run, to me, means they have all the time in the world to find the clears they want.
The other thing that others were saying last night was referring to the style of gameplay vs. other top level runners. This, perhaps, is the most subjective view I have with the least amount of concrete evidence. The raw gameplay within both the pre-established RTA and TAS videos is very consistent with how I approach these games. I do a lot of chaining through empty vertical columns, and pair it later with horizontal clears with limited amounts of lag chaining. However, through my involvement with the community over many years, I personally know that this strategy I use is not optimal in a speedrunning context for marathon mode. For speed, you need more lag chains and preparation, and less "moving blocks in at the last second" kind of strats. The fact that this runner can move a block in 2 or 3 times in a row, and then get exactly where they want over and again just seems too suspicious for me to believe it is real.
I believe this run, at the very least, should be removed from the leaderboards and have its verification status taken away until there is proof in a series of PB's over the months can show less mistakes and more speed with decreasing times. Until then, I believe that this is a TAS run trying to be passed off as an RTA run, which is a shame.
Under this run they explain their playstyle as such:
"I been practicing a few controller tricks in secret and after 3 months of practicing them, I finally got them down. A-B Spinning, Stick Controls for the SNES controller with the Sticks and Hy-birding between Rolling and Hyper Tapping as well as playing Panel Attack on Level 11 (EX-Mode) made grabbing this record and making it go this low possible."
I'm a keyboard player so this sounds like chinese to me, I don't know whenever these techs from other games would translate well to panel de pon. From the limited research i did into it, not sure how you'd alternate those two techniques (tetris rolling involves a weird hand position, so hand cam would be nice)
It seems pretty obvious to me that the submitted WR is TAS'd. All of the movements are way too fast and perfect, especially compared to Syuga's run of the game, who was the previous holder of the record.
As someone who has speed ran the 99,999k ex speed run category for "panel attack" 2years and have only had 10 successful runs from 3hrs sloly increasing to ~> 38mins and 40 seconds with major amounts of set ups. But also trying to find chains since I'm blind I can tell this is a TAS run because his movements for finding chains makes no sense. His video of ex true is (the category I run where you start at the given time) people have seen me attempt this in Real Time from the PA community. I Also am the only one with a submitted / verified run on a old user name which I still haven't submitted any other run cause I wasn't happy with my results and I've been playing ex all the time in Vs mode vs the community I'd also like to mention my first speed run with a full completed run which took 2 months to finish was grinded for over 12-18hrs a day and was beating by a person who only played on 6-8 speed magically become a ex master yet couldn't win a match against me I find a 7 min (true ex impossible todo unless your a GM player) <- at the time shippai was silver 10 months ago when submitting a 7 min run for PA Ex this run was also submitted 2 days after my intetional 1h 47 min
The key points is this is shippai - just going by another alias's he uses TAS tools and has stated he's used "Cheat engine" to help build TAS setups.
The other thing i'd like to point out from anyone in. 2min time aka Yoshi_aus_100 A grandmaster player who actively runs TA and some other top speed runners use and locate panels closet to where the chain is going to land this how ever is not using DAS which in order to get from where the current run noted from those videos with how the speed of hard is.
I seen some of his other videos discriptions he claims to have a sub 2m15s to 2m35sec run for ex in PA which truly isn't plausible this run the only person who's ran ex and 99,999 score cap in such minimal time is Yoshi_aus_100 , Myself and Z26. I can verfie Z26's 17mins and xx Seconds from like 1hour and 40 min acutal run times cause of how much he grinds ex like me and we show each others streams. His "true ex run" his APM just gradually stays above 500-600 which is extremely hard to do cause you have to manually find time to make chains which isn't possibly accurate from how he originally played shosoul have verified my run before.
The problem is the inputs shown at 30secs into the video shows him actively hitting left right left left right lift right repeatedly as a robot would with in a small time frame even on .25 secs almost instantly but the cusor moves faster based on specific areas and specific times. But slows down to submit false player intentions of submitting a legitimate run to seem like it's a world record when a normal player doing a hard mode run struggles more due to the speed of hard I still struggle on hard but not as much as ex since I actively play at a faster pace.
The other thing is that as a person who manually has to find combos and z26s gameplay we both actively try to manually and make chains before they actually go off and with high amounts of APM the cursor should be inconsistent. But actively made a RTA - for (7 min ex) and making TAS (1min 37s) runs on ex match the same amount of movement cause making constantly 10-13 chains with no setupis actively impossible on ex and actively raise I have made only a few 12-13 chains my intirity of ex and I grind for a good possible run I've had already 3000 attempts total and over 24,000 ex vs 10 match's for panel attack.
Z26 is a high play player who started playing ex after me and I can see his improvements, and say legit runs my skill rank vary's from copper, to plat. The other parts about those suppose low time runs just don't add up to his original game play, and have been constantly lowering older runs from his other user name "Shippai" and "Rita". But the amount of inconsistence's he throws actively in to make it look like it's legit but it's definitely not from the movement of human reactively acting to the inputs provided in certain points aka TAS. The speed for hard and ex are completely different ball games for timing/frames hover time and normal standard play "inserts are the biggest struggle"
Keynote - This user also actively gets mad if something doesn't go his way in any speed run he does
(my apologies had the wrong yoshci with Yoshi100_Aus) Yoschi Is a grandmaster yoshi100_aus is gold I got them mixed up my apologies. So many people with silimar names I do mix them up at times.
Thank you for doing such an in-depth analysis on this matter Cards. I am going to keep my thoughts brief: A full ban is appropriate as this run is a TAS.
I'm assuming raidabelow has been personally contacted and asked for his side of the story? Obviously he might just have chosen not to comment given the provided evidence, but he does deserve a chance to defend himself.
Since they are not part of the community Discord, I have had to create this thread and message them here to give them that opportunity.
i hear there are some that use the time slow ability, basically playing a game in slow mo, maybe 40-50% speed. perhaps that'd explain the fast inputs at full speed? doing that just makes it unfair to all the legit players who grind their butts off in a lot of games. being good at anything requires skill, not cheating.
I'm not super-active in the community, so I'm not going to make any suggestions on punishment. However, I've been playing Tetris Attack for 27 years and there's plenty suspicious here.
I wanted to point out a couple timestamps that weren't mentioned above. These both exhibit weird stack management behavior. They appear to me as either deliberate "errors" or the type of moves that would be made by a souped up COM player (which would be a cool creation but is not what these leaderboards are for).
Timestamp 1:13 – Clearing the short stack here with a x3 is very strange. The dark blue triangles are pulled from opposite sides of the well, and the chain is made deliberately. The normal strategy here would be to start raising immediately and downstack without clears.
Timestamp 2:15 – Setting up the x2 with the yellow stars while the light blue bonus clears is also strange low stack strategy. Timing the switch of the dark blue triangles to hit x2 6 and then continuing to x3 is nonsensical, especially since it still requires downstacking the star on the left side of the well.
I'd be curious if we can get them to play on stream or in some fashion we can see it live...I'd be willing to bet they don't play as well or the same OR will avoid the request all together... either way we have some kind of answer...
I agree it's sus... lots of inconsistencies for the level they're playing at. I think the ones I noticed are all mentioned... the biggest was the x15 or 16.. Also, there were a lot of times where I don't think any of teh top pvp or speed player would have made the movement that they did. There was a lot of waiting then going super fast instead of continuous work to set things up... It all feels too robotic to me.
Well he claims using ex speed from panel attack helped him get that speed, which ex runs 1.5x hard mode and wouldn't give you a advantage of in PDP cause lvl 9 speed aka (hard speed) (ex speed 11) faster then 9 speed if you attempt and far less frame data but he does this for more then one setting from easy normal and hard but is always using TAS to make his movements smoother if anything he's probly loading the same save state to studie it as well since it would load the same panel patterns (he also can't prove his runs in ex cause his true ex run) ways runs at 500+APM through out a 7 min which this is highly unlucky cause your fingers would get tired and constantly be swapping properly with raising as well but his other pdp runs show the same static movements and the pathing for chains and combos are not ideal for top lvl gameplay which is where the TAS comes in (he also only shows replays for his runs compared to showing any actual gameplay its self) the major points to look at cursor speed is above higher level players is robotic/ to fast even with hyper taping or rolling wouldn't be effective.
As cards stated raising on R or L perfect frames to raise makes zero sense considering you'd get the same effect from hold raising. (the reason I believe he was doing this was so he could effectively read the board and determine the TAS set up he wants considering how fast he's hitting both directionals even on (.25xspeed YouTube video the 25 sec mark up to about 36secs) this is not something a speed running world record player would struggle on unless a input was failed and or hardware issues but would still be easy to fix. I can personally say from my own PDP/TA runs on snex's emulator runs pretty jank at times. On original hardware my buttons sometimes stick so inputs can be lost as well.
But even my hard runs probly take about 30-40mins but I do all my runs recording myself with my game pad showing in my hands the controller / feed back to show in-game input l also live stream the game so people can watch and comment on it at the same time I might respond if I see chat and have my controller be able to be picked up by my microphone so viewers could hear said buttons being pressed. This is the easiest way for my self to show my verified runs / people playing on twitch shows a high creditability to themselves especially being a high lvl player cause you have viewers actually watch you attempt it but you can play a video make it look legit and just boom bam instantly click a hotkey to get live split to go off at the start to make it look like your actually running it as well. And I will say the more people are watching me the more nervous I personally get from speedrunning but I will still attempt to get a good run if not a avg run. I don't know his reasoning but I'm sure it's cause he wants fame and a title slapped on himself to boost his ego?.
