I also know a lot of runners that run games without using splits. Either because the game is too short (Caesars Palace any%) or because the game has IGT, or because they simply don't care for them (@MelonSlice)

Splits are great supplemental information, but aren't of much inherent value.

While this is not exactly on-topic, splits CAN serve a crucial purpose if they accompany video footage, as they can help you identify frame drops or encoding lag, which could otherwise go unnoticed and result in inaccurate timing of the run. I'd always recommend having splits (or generally just a timer) in your submission. You don't have to look at them during the run or even do anything, except starting them. It's additional information that can help others assess the footage.

The only instance I can see frame drops affecting accurate retimings is when there are frames drops on the very moment that you start timing or stop timing for the run. Even then it's not like the split timer will necessarily prove useful as there is no way to verify that the timer was started or stopped accurately in the first place anyways. I recommend splits for anyone that cares to learn what segments they are losing time on but it's not anything that's really too useful to the verification of the speedrun. I think timer's that are added post are the most useful way to visually verify that the timer was started and ended accurately.

If I were to see a run without a video, I'd just want all the supplemental evidence possible in order to feel more comfortable accepting it. It's less that providing splits means a run is legitimate and more that the inability to provide anything is fairly sketchy. Splits, a picture of the end of the run, really anything to show that you care about a run being perceived as real would help. Not being able to record video in this day and age is such a shoddy excuse in this day and age already, not putting in any effort to prove yourself on top of that is just unacceptable.

"The only instance I can see frame drops affecting accurate retimings is when there are frames drops on the very moment that you start timing or stop timing for the run. Even then it's not like the split timer will necessarily prove useful as there is no way to verify that the timer was started or stopped accurately in the first place anyways."

@Twan_Jones You're missing something crucial. Depending on the recording method and device, dropped frames are actually missing from the recording (my old Dazzle DVC100 has that issue, e.g.). That means that you would re-time an actual RTA second @30fps with 1 dropped frame as 0.967s, e.g. On the other hand, encoding lag (or buffering) can add duplicate frames to the recording - which can be impossible to tell apart from the game itself lagging just by the footage alone. In both instances, counting frames and comparing to the time span that has passed on the on-screen timer can help you identify such cases.