You might want to keep looking for your smarter people, if they believe QoS, optimally done is generally ineffective.
I do agree if an interface doesn't ever have a queue, QoS generally isn't needed, but in real networks, rarely are interfaces not oversubscribed.
I only very briefly skimmed your referenced video, and (perhaps) saw one of the biggest common mistakes, where they showed a link loaded at (?) 10%, 50% and 90% had no congestion issues. (I didn't see their method of loading the link.)
Firstly, understand an interface is always being used as just two usage percentages, 0 and 100. Anything else is a capacity average over some time period.
For example, if I transmit, continously for 10 seconds, and if your measurement period is the same 10 seconds, usage would be seen as 100%. If those 10 seconds were measured across 20 seconds, then usage would be 50%. But, even in the latter, for 10 seconds of the 20, link was completely saturated.
But, say for that 10 second transmission, a faster or multiple interfaces were feeding the traffic to an egress interface, and it couldn't queue more than 10 seconds worth of traffic. Say, actually 11 seconds worth of traffic was received. I.e. we dropped about 10% of the traffic. Yet, looking at just link loading for 20 seconds, usage was (only) 50%.
The "poster child" for QoS is mixing something like FTP and VoIP traffic on the same intrface, because FTP can cause transient problems for VoIP during the 10 second period.
Further, even if you were running just FTP, personally, I believe QoS needs to insure the best "goodput" transmission rate for the FTP traffic.
Don't misunderstand, I don't propose QoS is always of benefit, or always needed, but in my experience, its needed much more than many think.
Also, except for things like VoIP traffic, you typical class book QoS model, like Cisco's 12 class model, or the RFC model, I find often don't work very well in the "real world". Since they don't, unfortunally it furthers the common misception that QoS isn't very effective.
Personally, I wouldn't qualifiy myself as a smarter person, but referencing much QoS literature, I spent about a decade working to improve perceived network performance across an Americas WAN. Over much time, I found what QoS techniques were truly a QoS benefit, and what weren't.
Two stories I like to tell about the forgoing. First, the senior WAN engineer felt QoS was just worthless "voodo", well until the day he came to work and his phone was lit up like a Christmas Tree with people complaining "what's wrong with the network". Turned out, overnight one of our HQ routers rebooted itself and dropped it QoS settings. He restored the QoS settings, phone calls stopped.
Company I was doing this for, had 3 major regions, Americas, EMEA and Asia-Pacific. During one of their annual network conferences, two other regions were complaining, no matter how much bandwidth they purchase, poor network performance complaints just don't stop. I mentioned, since we started to use QoS, we seldom increase bandwidths, yet our uses no longer complain. Well, other regions also seemed to consider QoS "voodo" and continued as they did.
