Where does one post questions about possible VSX errors, such as discrepant positions? I thought I might find it in a VSX faq, if one existed.
There is no VSX
There is no VSX faq, but if you find discrepant positions you can revise the entry in question (log in to VSX and click the "Revise" link on the Detail Sheet of the object). After moderation your revision will then be promoted to the live database.
I tried to revise several parameters of couple of stars on VSX. Sebastian, patiently, warned me many times how to do that but I see that it is not an easy task for an amateur. If there was a guideline on how to revise each field with the minimum requirements, that would be very helpful for newbies like me. I see that there are too many things that can be done, actually needed to be done, with data mining for VSX. But it appeared to me that it is not that easy task if you are not familiar with the requirements or the steps needed to change even just one field. May be, AAVSO should consider to create a guidebook on how to revise VSX database. OR much better would be to open a CHOICE course on that. I would definitely sign up for that. I think "data mining for revision" is as important as submitting observations to the database.
Anyway, I have one question: Suppose I entered a lightcurve (including minima for an EA star, for example) to the database and it is the only data in the database. Moreover, the VSX epoch for this star is too old. I want to revise the epoch with my data. Can I do that? If so, how?
Thank you in advance.
That guidebook exists and we have extensively publicized it.
We have a manual, several help pages and also the forms have examples written along each field.
I usually give these links to all VSX contributors in case they do not know about them:
The manual shows detailed examples of submissions and revisions.
About your EA question. The most likely case (no period change involved) is that you can improve the period so your minimum and the published epoch fit. It is better not to change the epoch but to improve the period.
I don't know exactly what you mean with "the only data in the database". In which database? VSX? AID? All surveys? If there is already an epoch in VSX, then the best choice is always using all the survey data available to get long term coverage and combine them in a single phase plot. The old epoch will most of the times be useful to improve the period even more (longer baseline).
If you only have new data, e.g.; only spanning a few night, it is not woth to submit a revision because there is nothing that may indicate that the new period is better (the old one might be based on a longer time span). The recent epoch would be more useful for predictions but we would be losing an old epoch that would be useful to refine the period. Combining the old epoch with a new epoch if it is based on only a single night or a few nights wouldn't be useful either because there would be a cycle count ambiguity.
Patrick already replied but I'd like to add that if you improve a star's position, always use Gaia DR2 (J2000.0) coordinates as given in VizieR or SIMBAD.