Sometimes you absolutely need to work with a particular source which may be an absolute pain to work with.
I'll start with a couple of examples, but what about others...do you have a love/hate relationship with a particular source?
Around 200 years ago in my country (Iceland), when priests were supposed to keep detailed and accurate records of baptisms, confirmations, marriages and funerals, as well as making a parish census every 2 years, there was one priest with a problem - a pretty severe alcoholism problem, it seems. He even got reprimanded for giving a sermon on one occasion while being barely able to stand upright. Unsurprisingly, his records are a total mess - a number of entries seem missing or badly garbled, wrong names, or next to illegible writing. What makes things worse is that there was no government census in this area between 1801 and 1835, and this was a fishing community, with a lot of people from elsewhere, seasonal workers (which did not stop children from being born, of course) and so on, so his records, as bad as they are, are still the main source for this parish during this timeframe. Whenever I need to look up someone from that region during this period, I cannot help but groan just a little bit.
Then there is the census of 1762. There was a full census in my country in 1703 and another in 1801, but in between there are mostly just partial censuses. Then we have the 1762 census, which should be able to fill in some gaps, but there are just so many problems with it.
First, the authorities who sent out the forms did not include instructions for the census takers, so a number of things got interpreted in different ways in the different counties. In some of them, just the heads of the household are listed - in other cases everyone. Sometimes the relationships between individuals are listed ("his mother", "his/her/their children", "his wife's sister" and so on), and sometimes not. Then there are the ages. There are 4 different columns for the names of individuals, and 7 different columns for their ages. Now, sometimes things are clear, with one name, and one (corresponding) age per line, but you can also find multiple names per line, and maybe the ages are just listed in decreasing order, with no way to match a name to an age.
Ugh.
There are even some more bizarre problems with it. At the time the census was taken, my country (Iceland) was effectively a colony of Denmark, and the census was done for the Danish authorities, and supposed to be in Danish. Now, one problem with that is that Icelandic and Danish do not use the same alphabet. Danish has two letters that Icelandic does not use, and Icelandic has 9 letters not used in Danish.
Now imagine what that does to the transcribing of the names.
Ugh again.