Rotten Raters Я Us

Monthly TRP Graphs and the BS we get from our overloads

Quick Links to Monthly TRP Results

November 11, 2025 | October 14, 2025 | September 16, 2025 | August 19, 2025 | July 22, 2025 | June 24, 2025 | May 27, 2025 | April 29, 2025

---

November Factuality and Severity TRP Graphs

Results from November 11, 2025

---

October Factuality and Severity TRP Graphs

Results from October 14, 2025

While we're all aware that the Wayback Machine is mentioned in GG as a reputable place to verify the information, it's apparently the end-all as well. The query for Mexican Restaurants in Detroit, had a response date of March 27, 2025. And since the menu for Evie's Tamales in Detroit wasn't available on the Wayback Machine, this was CCA. It's almost like this was a trick to intentionally screw with us.

---

September Factuality and Severity TRP Graphs

Results from September 16, 2025

---

August Factuality and Severity TRP Graphs

Results from August 19, 2025

---

July Factuality and Severity TRP Graphs

Results from July 22, 2025

Let's get to it, eh. On the Chinese restaurant task we learned not to be literal and to use the context in the response, even though the target sentence is clearly wrong. There is a complete lack of consistency, kind of like Mamdani changing his views on a whim. And we learned that Indiana and Kentucky are the same state, in'nit? There was no hedging in the Kentucky severity task, but they still accepted the Ohio Falls State Park as a good suggestion due to it's proximity to Kentucky, and a reasonable suggestion. Yet, a few months ago with the HP Printer severity, they absolutely refused to consider 27.78 and 27.8 as the same weight. There is no consistency with their ratings on these. The other task which doesn't make sense is Simone Biles. We were told this is a clear-cut inaccuracy, yet the rating only included the low ranges. The medal count that was given was completely wrong and stale, yet they are ignoring this fact. However, they are not ignoring the part about awards in the AMA Awards severity, here it was considered a strict high. How is this more harmful than the stale and inaccurate medal account for Simone?

---

June Factuality and Severity TRP Graphs

Results from June 24, 2025

They cannot give us consistent guidence on context. This month the problem task was [rp diet coach app cost]. The evidence specifically said "Monthly Subscription: $14.99 or $19.99 per month." This came directly from this header, "According to Sensor Tower, the app's subscription costs are." So one would think you just check the Sensor Tower website and verify, but no that is not correct. This is what we were told Sensor Tower lists several different prices though it is unclear if these are current costs or previous promotional deals. We cannot tell from the website if both prices are currently valid and if they are, what factors affect which price a user might pay. However, the official site for the app does not mention a $14.99 monthly subscription fee. Only the $19.99 subscription fee is mentioned. The claim is marked as inaccurate due to misleading information." So now context isn't enough, we have to verify the context information.

---

May Factuality TRP Graph

Results from May 27, 2025

---

April Factuality and Severity TRP Graphs

Results from April 29, 2025

The insanity for this month is the email distribution list fed by a Google Sheet. We were told the inaccuracy here is somewhat related to primary user intent. There is a mitigating nuance and context when reading the full instructions for duplicate management in Google Contacts. It is automatically done after you select the option to do it. This makes the inaccuracy of the claim in context less severe, which results in the severity rating of low, so this is a strict low. What regular person is going to care about that "mitigating nuance"? Perfect example of how we don't stand a chance to do well with these subjective tasks.