Take perspective of the modeller. What would be the first circle of influence in teen suicide? Probably direct social interactions (family, peers). Second circle? More abstract things like succes in school, relationships, the absence of bullying. For a regulatory change to have a meaningfull and measurable effect on suicides (literally the most drastic step available) is hard on first sight.
And then the step to research design. How would one expect to find a determinating factor by a DiD-design with only some regulatory measure as an explanatory factor?
For me, after looking into this study. First explain me the state by state differences in teen suicides over time adequately, before you start explaining the effect of gay marriage on teen suicides.
By the way, if you look at the accompanying figure 2 and figure 3 in the JAMA article, you will see just how noisy the data is for both Wave 1 and Wave 2 states, as well as the relative lack of post-2014 data points. Honestly, just looking at those two charts should be enough to give you a healthy skepticism about the conclusion, whatever your ideological views.
Along with 'gay marriage', 'gay acceptance' has been on the rise in general. Social interventions, heck, social media and the internet (i.e people can form like minded networks etc..).
Even a small change in policing and social policy could cause this. A bullying program.
There are so many things that could 'cause' a tiny drop in teen suicide.
Statistical noise is always present, so of course suicide among GSM teens did not uniformly drop.
The JAMA study used different years for different states, as same-sex marriage became legal in different states at different times. His criticism about "any event in 2014 explains this drop in suicide" is malformed.