Here's the problem with gun torture tests - it's generally whatever the tester feels like doing during the test and not a fixed, formal set of procedures developed to actually impartially test the gun.
I wrote test procedures for a number of years for a variety of clients. What you try to do with any type of testing is setup a series of tests that:
1. Subject the equipment being tested to the same procedures under the same conditions.
2. Every test procedure should have a reason for being done so that a specific performance requirement or feature is being demonstrated and tested.
3. Test procedures have to be repeatable so that the item being tested can be retested as required.
4. When possible, test procedures should not require special knowledge or expertise and can be done by anyone by simply following the procedures.
5. Tests should be repeated multiple times to ensure the result (pass / fail) is the real outcome and not a one-time anomaly.
You want to test a gun - then setup a series of procedures that include the same testing for every gun. Write them down with a description of the function or feature being tested and how the procedure will prove the function or feature with the test.
If you want to test a gun for water, mud etc. - it's not that hard to do. You get a series of 5 gallon buckets and start with clear water. The next bucket has an exact mixture of water + the sand + dirt of your choice added to the water by weight and stirred just before putting the gun into the bucket so that everything is in suspension when the gun is immersed in the test medium. In effect, you have a "calibrated medium" that you can make whenever you want to test a gun. It has a main testing feature - it's repeatable at any time.
Immerse the gun in the medium unloaded, slide locked back or chamber open, then load the gun and test for function. Then completely clean the gun, lubricate it, test it for function and repeat the test again. Clean, lube, function test, and repeat again.
If you want, make multiple buckets with progressively thicker mixtures. You can incrementally increase the thickness of the water / soil / sand mixture until it's mud - if that's what you want to do.
Then repeat the test at least three times in each bucket to ensure you are getting the same result and that the first one wasn't a fluke.
Then repeat all of the testing again, but this time with the gun fully loaded (round in the chamber + loaded magazine).
Yeah - it's a lot of boring, repetitive work. That's exactly what real testing is - controlled, repeated function tests because that's how you prove functionality
What testing doesn't include is freestyle, unrepeatable running around a site and throwing the gun at whatever strikes your fancy at the moment. Real testing has to be controlled and repeatable to be worth anything.
But, what I see is some guy in a totally uncontrolled environment, subject a gun to variable conditions that can never be repeated - and then making a failure claim in comparison to another gun tested under completely different conditions.
That's not testing. That's simply screwing around on video and then posting it on YouTube and making whatever claims you feel are sensational enough to get as many viewers as possible. It's simply a joke and laughable at best. It's not even remotely related to real testing.
MattV2099 tests are just as realistic as any other testing - he's just being outrageous and about it and spoofing "gun tests" as being ridiculous - because, for the most part, they are. However, one thing about his tests - they can be repeated by anyone at any time to verify the results he's showing.
Then you have people like Todd Green, who simply got a gun and used it for a year noting the performance and failures over a certain number of rounds. He usually shot about 60K rounds as a minimum through the gun, and in the case of the HK VP9 about 90,000 until he had a catastrophic failure that would have caused the gun to go back to HK to be fixed.
What's more realistic? A guy running around doing whatever seems to strike him at the moment as a "test" - or, a guy who takes a gun and shoots it for a year under his normal shooting regimen including both practice and self defense / pistol classes, and competition?
Probably neither - as both are uncontrolled and not repeatable - but, people will come away with whatever piece of information fits their personal agenda in relation to the manufacturer and product instead of any real, meaningful data.