Primary and Secondary Sources -- Who do you trust?

All categorizations of sources are arbitrary. Trying to pigeon hole any source as either "primary," "secondary" or some other shade of classification can obscure, by the very nature of classification, the real value or lack of value of the source. Let's say I made a decision to go out and only look at "primary" sources. First of all, where is the consistency in the definition of primary vs. anything else? Here is a list of possible sources. Which of these is primary and which are not and would be considered in some other category?

U.S. Federal Census Record
Official State Birth Certificate
Official State Delayed Birth Certificate
International Genealogical Index
Ancestral File
Ancestry.com Family Tree
Church Parish Register
County Assessors Recordings
City Tax Roll

OK, that's likely enough examples. Can you positively say that any of these named types of documents or entities did not have primary source material (as you understand the definition) Perhaps we could review the common definition of primary source: that is, a document prepared at or near the time of the event by someone who actually participated in or witnessed the event." From looking at this list, are you prepared to say which of these are or are not primary sources? Or may or may not have primary source information?

If you look at a few dozen family group records or individual records from a sampling of researchers, you will find every single one of the items on the list above used as a "source" for some event or the identity of some person. So what am I really trying to say here? As a matter of fact, following the definition I just gave of a primary source, every one of the items on the list could possibly have information in that category. What I am saying is that learning and applying discrete categories of "evidence" obscures the process of determining the value of the documents. Rather than applying an ad hoc label, you should be mentally asking a series of questions about the evidence, similar to those given to jurors at a trial. Here is a sample jury instruction:

A witness is a person who has knowledge related to this case. You
will have to decide whether you believe each witness and how
important each witness’s testimony is to the case. You may believe
all, part, or none of a witness’s testimony.
In deciding whether to believe a witness’s testimony, you may
consider, among other factors, the following:
(a) How well did the witness see, hear, or otherwise sense what
he or she described in court?
(b) How well did the witness remember and describe what
happened?
(c) How did the witness look, act, and speak while testifying?
(d) Did the witness have any reason to say something that was
not true? Did the witness show any bias or prejudice? Did
the witness have a personal relationship with any of the
parties involved in the case? Does the witness have a
personal stake in how this case is decided?
(e) What was the witness’s attitude toward this case or about
giving testimony?
Sometimes a witness may say something that is not consistent with
something else he or she said. Sometimes different witnesses will
give different versions of what happened. People often forget
things or make mistakes in what they remember. Also, two people
may see the same event but remember it differently. You may
consider these differences, but do not decide that testimony is
untrue just because it differs from other testimony.
However, if you decide that a witness has deliberately testified
untruthfully about something important, you may choose not to
believe anything that witness said. On the other hand, if you think
the witness testified untruthfully about some things but told the
truth about others, you may accept the part you think is true and
ignore the rest.
Do not make any decision simply because there were more
witnesses on one side than on the other. If you believe it is true,
the testimony of a single witness is enough to prove a fact.
You must not be biased in favor of or against any witness because
of his or her disability, gender, race, religion, ethnicity, sexual
orientation, age, national origin, [or] socioeconomic status[, or
[insert any other impermissible form of bias]].

This sounds like a pretty good way to evaluate evidence without getting into the bugaboo of classification of the evidence into discrete categories. Of course, we usually do not have benefit in genealogy of actually hearing the testimony or seeing the witnesses as they gave the information, but you can use the rest of this instruction as a method to weigh the evidence. We also need to remember that within the same document or record part of the entries may be trustworthy and others may not. Every piece of evidence needs to evaluated on the basis of its consistency, historical context, timeliness and believability.