deletions | additions       diff --git a/section_Conclusions_and_Outlook_xx__.tex b/section_Conclusions_and_Outlook_xx__.tex  index eff27c9..1b166ff 100644  --- a/section_Conclusions_and_Outlook_xx__.tex  +++ b/section_Conclusions_and_Outlook_xx__.tex 
...
While our study identifies cases where semiempirical methods give results that differ significantly from the DFT and may require further attention, it is clear that five systems is not sufficient for a general and statistically significant assessment of the accuracy of a computational method. We plan to add more systems from the literature to the data set and invite other researchers to do the same. This can easily be done by (1) making a free account on Github, (2) making a pull-request ...  While B3LYP/6-311+G(2d,2p)[LANL2DZ]//B3LYP/6-31G(d,p) may be an adequate level of theory to identify deficiencies in semiempirical methods it is unlikely to be accurate enough to parameterize against. In the case of intermolecular interactions the "gold standard" is CCSD(T)/CBS//MP2/TZVP computed using extrapolation \cite{Jure_ka_2006,_ez__2013}. This level of theory may be impractical for these size systems for the foreseeable future, but could be approximated by extrapolating from smaller systems using an ONIOM-like approach (xx) \cite{Chung_2015}  or approaches like DLPNO-CCSD(T) \cite{Liakos_2015}. Again we encourage researchers interested in developing or testing such methods to use the coordinates in the dataset and deposit the barriers and reaction energies.