“It’s me again. Once more I’m here to cut down the time almost in half.” With those words, Groobo introduced the spಞeedrunnꩵing world to his 2009 Diablo record, the fastest Diablo run the world would ever see.
In 2004, he beat the previous best time of 53 minutes with a staggering 17:38. In only five years, he brought the record down to 3 minutes, 12 seconds.
No one has taken over the record since. N💎o one ever overtook any of Groobo’s times. The Diablo speedrunning community moved away from the unbeatable Groobo runs, retreating to the increasingly popular categories of single segment, no negative HP glitch. 15 years later, G🌳roobo would be accused of cheating, and all his Diablo records would be scrubbed from leaderboards.
How spliced runs work

Even though Groobo’s run is much shorter than any previous attempts, he divided his video proof into segments. The previous record holder had taken 5 separate videos to record their 53-minute-long attempt, making each video about 10 minutes long. Uploading videos to the internet wasn’t easy in 2004, so this was nothing out of the ordinary. Meanwhile, Groobo’s first record of 17:38 was divided into 19 sections, less than one minute per video on average, which grew to 40 sections as his time shrunk to 06:07. His 2009 run was just 03:12 long, but it was delivered in 27 videos, each only 7 seconds long.
Was this not suspicious? Couldn’t Groobo have optimized his run by retrying each 7-second-long section until he had done each perfectly? Why, of course he did. That’s what a segmented run is. Most challengers might not have used as many segments as he did, but a high number of slices a💯lone is not enough to raise the alarm.
As the said in 2009, before being edited: “Use as many segments as is optimal to achieve the fastest final time. We will not be more impressed if you use a small number of segments.”
Splicing rules used to be quite loose, too, clearing Groobo of what would otherwise be cheating. You can see something wrong in the version of Diablo used for the world record before the run even begins. The copyright notice on Diablo’s splash screen reads “Copyright 1996-2001”, which would maဣke this version 1.09. It certainly isn’t the 1.0, or it would just say 1996.
Meanwhile, the main menu on the next screen calls itself Diablo v1.00. Groobo used two different versions of the game, and yet he wasn’t disqualified. SDA’s rules state clearly that runs are allowed to “have discrepancies between two segments as long as none of it is to your advantage”. The run only begins when Groobo s♑tarts a new game, and we’re still on the main menu, so this isn’t cheating.
Suspicions of illegal splicing

There are multiple instances of Groobo not splicing his 2009 run as he should have. To be fair, instructions on how to splice runs are vague, but it’s clear that he didn’t act as he was expected to. The worst of🦄fender by far is the disappearing magical ring, which is placed in the inventory before vanishing without a trace, only to rea🐎ppear a few seconds later. But again, this might not be against the rules because the discrepancy doesn’t help the runner in any way. But even if Groobo didn’t need that ring, its disappearing act indicates some serious meddling with the sections.
Those suspicions might have led to the realization that the (very convenient) map layout found in Groobo’s record must be the result of editing together different runs with different seeds. Even if they did, proving this is extremely complex because it requires retro-engineering the way the game generates its map and distributes the s𝓡ide quests.
These inconsistencies can be inferred by the shape of the randomly generated map, but they can’t be intuitively understood by players, not even other runners. Extremely convenient stairs and items did raise the suspicion of cheating, but finding definitive proof meant more work than anyone was willing to put into debunking an old speedrun record.
Evidence of multiple Diablo versions

Unlike dungeon layouts and improper splicing, evidence that Groobo used multiple versions of Diablo is not hard to find, if you know the game well🗹 enough. Here’s the short version.
Groobo uses a lot of glitches in his run, but only two stand out: item duplication (dupe) and negative HP. The negative HP bug is pretty much banned in modern Diablo speedrunning while duping꧃ is important to bridge the two biggest categories, Any% and Any% No Dupe.
The duplication glitch became m🍒uch easier to perform starting with version 1.02. Not easy, but easier. Performing it multiple times in a row, as Groobo does, is so unlikely to be basically impossible in version 1.00.
At the same time, the negative HP glitch was removed in version 1.07. Thi😼nking back to the main menu inconsistencies between version 1.00 and version 1.09, you might realize something. 1.00 is smaller than 1.02 and 1.09 is higher than 1.07. The first is incompatible with item duplica🍌tion while the second doesn’t feature the negative HP glitch.
If Groobo was running not 1.00 nor 1.09 but on a third version of the game, why were both versions in the first two screens of the run? The only reasonable answer is that the whole playthrough, not just the splash screen and the main menu, is based on multiple runs of multiple versions of Diablo. That, surely, must prove cheating.
Why was Groobo not caught immediately

None of this proves cheating. Sure, it strongly suggests the possib𓆉ility of cheating, and if the 2009 record had been set by𒁏 a no-name first runner, it would have raised a lot more attention.
Proving that the run was impossible required the to find that no seed could recreate the run’s procedurally generated map. No one is going to do that much work to debunk the recordဣ of a somewhat well-known speedrunner who was, ultimat🦹ely, a prolific member of a small community.
That’s the reason it took 15 years to catch Groobo. No on💫e wanted to look. The investigation only started when TASbot founder Allan Cecil began using Groobo’s 2009 record as a foundation for his tool-assisted speedrun. He was caught because a competitor looked at his record for inspiration. Because speedrunning is a sport, and no one watches the Olympics looking out for possible cheaters. No one likes to think their idols don’t deserve a place on the podium. And no one, especially not a volunteer-run website from 2009, wants to stop their rising star from celebrating a new victory.
Published: Feb 25, 2025 09:51 am