Respectfully, getting upset at the varying degrees of house damage being slapped with one rating is not the right takeaway here. That’s not inherently problematic, and the EF scale is actually specifically designed to allow for that to happen. The biggest issue here was the failure to consider or factor in tree and contextual damage that was not consistent with an EF2 tornado. This contextual evidence could have been used to justify a higher rating in Lake Village despite poor construction, but it wasn’t. But significant variance in destruction levels to multiple homes being classified under a single rating is something very commonly seen in tornadoes, and when it’s reasonably linked to variance in house type/construction quality, it is absolutely valid. EF2 damage can accurately encompass total destruction of modular/manufactored homes, and simultaneously accurately encompass well-built frame homes that have lost portions of their roofs (both types were hit by this particular tornado). On top of that, you have the difference between high-end and low-end EF2 damage, which are objectively two quite different types of damage in terms of physical appearance and should in no way be considered being “rated the same”. What I’m trying to say is that EF2 damage isn’t only valid if it’s the stereotypical house with its entire roof completely removed, nothing more nothing less. Same thing with EF3; a house with full hurricane code construction being left mostly intact, versus a largely unanchored old farm home being swept from its foundation are both totally valid under the same rating of EF3, but are totally different in appearance.
When it comes to rating house damage, the scale is specifically designed to NOT work on a “Just rate every home that is destroyed to a similar degree the same way every single time, based purely on the physical appearance of the damage” basis, and for good reason. In fact, this is exactly why the scale has established upper and lower bounds for each DOD.
Now where we start to have problems is when surveyors go way outside of those bounds without any reasonable justification, or when contextual evidence is not taken into account. In this particular case, it’s more the latter.