The contribution Imai, Lo and Olmstead article is... an algorithm. Not a model or technique or estimator, but a faster method for doing approximate inference.
So the APSR is publishing Machine Learning articles now?

LOL, hey we applied a 40 year old optimization method invented by some Harvard statisticians to "quickly" estimate parameters for "massive" data. Good stuff guys. Next they'll be telling us that they invented the field of convex optimization because they applied it to a political science problem.

APSR has always published articles like this. For example King's imputation algorithm. At least the Imai Lo Olmstead article isn't incorrect.
It's not general interest. It's of interest to a small group of scholars doing a particular type of work, and only the most technicallyinclined among them. It's a perfect fit for PA.
Imputation is of more general interest, even if King's work is junk.

The paper could have been published in a good stats journal. It is a fine paper, unlike a lot of Imai's and almost all of King's work. This isn't the paper to pick on if one wants to go at these people.
LOL, hey we applied a 40 year old optimization method invented by some Harvard statisticians to "quickly" estimate parameters for "massive" data. Good stuff guys. Next they'll be telling us that they invented the field of convex optimization because they applied it to a political science problem.

I disagree. All the authors have accomplished is applying variational EM, which has been around for several years now, to a specific class of ideal point models. This would never have made it into a stats journal, but would be a good fit for PA. I have no clue why the editors decided to publish it in the APSR since it has zero substantive relevance, this is our flagship journal for f*cks sake.
furthermore variational EM scales horribly and would be slow as all hell with actual large datasets so not really sure what the authors are selling here.
The paper could have been published in a good stats journal. It is a fine paper, unlike a lot of Imai's and almost all of King's work. This isn't the paper to pick on if one wants to go at these people.
LOL, hey we applied a 40 year old optimization method invented by some Harvard statisticians to "quickly" estimate parameters for "massive" data. Good stuff guys. Next they'll be telling us that they invented the field of convex optimization because they applied it to a political science problem.