Not impressed with the video. It appears to me that he had a desired outcome, and set up his tests accordingly.
First off, that small "unrated" (his words) shackle was never intended for this application and/or load. No point in testing it. However, he used is as a data point to make his argument. Worthless.
Then, he wraps the soft shackle around 2 hard (sharp) corners. In no application using rope/line does one ever wrap cordage around a 90* corner. That is just not done. As kids, we learned (by experience) that rope gets cut and fails when wrapped around a corner. By doing so, he shows either great ignorance or an intention of forcing an outcome. Or both.
Yes, he did re-run the test by wrapping the soft shackle around a radius, but that seemed like an afterthought. At least the soft shackle failed in an expected manner.
What is missing is showing the shackles in the "about to fail" state. The soft shackle gives more warning than the hard shackle before failure.
What you have here are two tools with the same primary function, but each with its own advantages and disadvantages in a given situation. By disallowing one, you lose all the advantages it has over the other.
The key is to make sure the rating of your tool is greater than the load you apply to it. For American made products, the SWL should be 20% of the failure load - a 5x safety factor. The same does not hold true for (some) off-shore manufactured products.
Use the right tool for the job.