Here again I would say that we "measure" supernova luminosities and redshifts, CMB photon redshifts etc. Then we fit a certain model to these measurements to determine model parameters, like lambda. Likewise, we do not observe acceleration, except within this certain mathematical framework (and indeed you can find some (unrealistic) models which fit the supernovae and CMB data (BAO gives enough tension to rule the simple models out) which are locally decelerating everywhere). In these models you would not have a cosmic event horizon at all.
I think the Lineweaver article is very reasonable, but I also felt like there should be a proviso that this relies heavily on the assumptions put into the standard cosmological model.