Oat Wii Legitimacy

This rationale is used by the SMS RTA Leaderboard Tracker.

Guy2308’s 1:13:17 and Oatflaker’s 1:14:56 pbs (as well as their 120 Shine pbs) were done on a specific Wii that gives fast loads, known as the Oat Wii. To summarise what’s happened here, the original claim that the Wii had an advantage that necessitated a hardware mod has been rejected as more evidence has emerged.

Original Claim

The 1:13:16 was achieved using a hard-modded Wii Family Edition disc drive. Those disc drives can not read GC discs without being hard-modded; resulting in faster loads.

– inkstarLum (2023/02/23 tweet)

Note that the 1:13:17 and 1:13:16 refer to the same run, whose time was changed following a change in the offset used for intro skip.

There were two major parts of supporting evidence used here:

  1. The total loading time of the 1:13:17 was (believed to be) less than any other known run done by runners other than Guy2308 and Oatflaker;
  2. The disc drive the Wii used is a rare revision that is used in family (edition) Wiis, which cannot read GameCube discs, and it was thought to not naturally occur in original Wiis.

On this evidence, I decided that the anomalous total loading time of the 1:13:17 meant that a decision should be made on balance of probabilities on whether this anomaly was possible on an unmodded Wii, and that it was likely modded because of:

The case for it being modded was weak, but the anomaly existing in itself merited leaning into this interpretation. Both runs were retimed to an example of a slow-loading run, which was JJsrl’s second-place 1:13:30, and so were adjusted by 19.025s. This was done to remove the advantage of this Wii over a random Wii.

In hindsight, I think I was looking for an explanation (anomaly = mod) where there wasn’t one, so want to take a more accountable approach to reviewing the evidence.

Evidence Review 1: Load Timing

Since the original claim and adjustment, I found a run (toobou’s 1:14:22 pb) that had faster loads – see timing. The results suggest the 1:13:17 has loads that are around halfway between the 1:14:22 and Weegee’s 1:14:13 (the latter is not fully retimed so this is a prediction).

This means the 1:13:17’s load total is no longer anomalous, so it changes the assessment of Oat Wii being modded from requiring balance of probabilities to requiring a strong case to merit rejecting a pb that had at least several hundreds of hours of work put into it. So the original case was rejected, and I reviewed the hardware evidence.

Evidence Review 2: Hardware

The disc drive revision under investigation is (informally) known as D4-2, or D4 with 1 IC (integrated circuit). Note that this versioning specifically refers to the drive’s chipset, and other parts of the drive are versioned independently. This drive was first identified by the central resource for Wii disc drives, www.wiidrives.com, from a photo submitted by SMS speedrunner Tareq of the Oat Wii itself. The bottom of its identification guide credits Tareq for the discovery of this revision and for a photo, and Tareq himself attests to submitting a photo of Oat Wii’s disc drive. Note that the photo on that webpage is spliced, and this seems to be the original.

Hence, Oat Wii has a D4-2 drive. This drive has been observed in family-edition Wiis, but the question is whether it is naturally extant in original-edition Wiis – or if it’s possible for it to be capable of reading GameCube discs without a hardware mod.

Following the original claim, inkstarLum shared evidence of the discovery of another Wii with a D4-2 drive. It has a drive photo and sample SMS video whose Delfino loads (sample of 12) were retimed to average 2.889, i.e. very close to Weegee’s 1:14:13. Note that nothing is known of the hardware of either toobou’s 1:14:22 Wii or Weegee’s 1:14:13 Wii.

This indicates that it is likely that drives exist that have a D4-2 chipset and can read GameCube discs, and so presumably appear in original-edition Wiis. It’s not certain because this newly-discovered Wii may also be hardware-modded.

However, experiments with family-edition Wiis indicate that their drives only differ from original-edition Wii drives by the addition of physical features that stop mini-discs from entering (source 1, 2), noting that all GameCube discs are mini and Wii discs are normal-size. In particular, modding a family-edition drive to accept GameCube discs requires only cutting off or bending one metal tab, or using a 3-inch disc adapter on the disc. It’s doubtful that such trivial hardware mods would even be considered illegal in a speedrun ruleset.

Beyond that, it would be unsurprising for hardware revisions of different Wii components (here, D4-2 chipset and the present/absent metal tab, or D4-2 chipset and Wii family/original-edition) to be mixed, given the muddy pattern of evolution of the drive chipset itself, as presented in the identification guide.

It is also unclear to what extent the D4-2 drive is responsible for fast loads. Both observed D4-2 drives yield loads that are much better than observed D4-1 drives (which average 2.94–2.98 Delfino loads), but there is a substantial difference between them (2.85 in the 1:13:17 vs 2.89 in the second discovered D4-2 Wii above). The difference is thought to be possibly due to laser wear or disc wear (tho I don’t have citations). Note that D4-1 is the previous known chipset revision chronologically, but may not be the fastest non-D4-2 revision.

So we conclude that the evidence suggests the drive does not gain an advantage from a ruleset-illegal mod. Therefore, the 1:13:17 and 1:14:56 (and any 120 Shine runs from these runners) are reinstated with unadjusted times on the leaderboard.

Other Clarifications

I’m sorry for claiming Oat’s Wii was modded with little supporting evidence. That was a bold claim to publish, and very misleading. I did believe the Wii was modded, however I was VERY likely wrong. I apologize to anybody who feels mislead, and the anonymized users.

– inkstarLum (2023/05/14 tweet)

inkstar has retracted the original claim.

Hey everyone!!! Official announcement incoming
Shoutouts to @InkstarLum doing research and confirming
I OFFICIALLY HAVE THE SMS ANY% WORLD RECORD
The anonymous run of 1:13:17 is confirmed to have used a modded wii to increase [sic] load times, which is banned
RTW for awareness!!

– JJsrl (2023/02/23 tweet)

It is not true, nor was it ever true, that Oat Wii was modded in order to reduce load times (it is unknown if it was modded with this intention, but it is believed to be very unlikely).

There is also no such notion as an official SMS Any% World Record. There is no universally-trusted authority on the world record or rankings in general, nor did the most popular leaderboard (speedrun.com) remove Guy2308’s pb for this reason (it was removed in 2021 for political reasons, that is, reasons unrelated to the run’s legitimacy). World record is objectively defined, but it will always be ambiguous which run it is owing to the unverifiability of things like hardware integrity. Thru assessing the available evidence via the case above, this leaderboard says that Guy2308 (and nobody else) has the SMS Any% WR (EDIT 2023/08: it has been beaten by a new run). This may change in future if the consensus on timing standards changes to remove loads for example.