I spotted a paper with around 106 citations, around which 86 are just the author citing himself. Now I understand that people do this to work on their previous studies, but 86? That's around just 20 citations you've received for your paper. Another paper of his has 96 citations with 87 being self. Is this normal or excessive?
Self-citations are valid, but too much is a bad sign:
- On the one hand, an author with an ongoing research program will generally have significant prior work that needs to be cited.
- On the other hand, all scientific work exists in a larger intellectual context for which Joy's law applies: lots of other smart people are doing related work that needs to be acknowledged.
High rates of self-citation thus generally indicate authors that are either:
- intellectually isolated,
- delusionally arrogant, or
- deliberately manipulating citation statistics.
But how high is too high?
As a rough heuristic, I have come to regard 1/3 as a useful upper bound on reasonable self-citation. That means that most citations are elsewhere, even for papers with very strong prior work to cite. I also generally find that the higher the total citation count, the lower the fraction that are self-citations (i.e., volume of prior work scales more slowly than complexity of scientific context). Conversely, the ratio may break down entirely in short works with artificially limited citation counts, e.g., in extended abstracts where all but a few absolutely mandatory citations are pruned for reasons of space.
Applying these heuristics to the cases that you describe, I would assess the citation ratios 86/106 and 87/96 as almost certainly indicating deliberate manipulation of citation statistics. If the fractions were similar but the citation count was small (e.g., <20 total), then it might be legitimate or a result of intellectual isolation. Even with lots of prior work, however, there is almost never a reason to cite so many different pieces of it, as a prolific author will generally also write review papers that are better to cite than the individual papers that went into them---and review papers have limited self-citation because they are describing other people's work as well. The author might be delusionally arrogant too, but there's a lot more citation manipulators out there than researchers who are both delusional and highly prolific.
Bottom line: probably deliberate citation manipulation.
Citations are neither a competition, nor a prize. They are simply a tool to refer to previous scientific work in order to establish the scientific case of the paper.
The fact that some people have decided that citations are in fact a good measure for "success", or a good case for "promotion", does not mean that citations should be perceived as anything else than a scientific tool, rather than a merit, prize or a favour.
Thus, there is nothing "excessive" in itself in loads of self-citations. Each citation should be evaluated solely based on the scientific justification of the cited work and its relevance to the current work, and there is nothing in your question that provides a witness for an unjustified citation. In other words, it is impossible to answer your question based solely on numbers and percentage of self-citations.
In other words, the question about many self-citations is whether the citations are relevant or not. Some authors work in a specialized field, or with specialized approaches in a popular field. So it is relevant to cite themselves. Overall, the problem of high percentage of self citations is different: if they are almost the only ones working in their subfield or using their approach, then maybe it's not a good approach?
There are certainly cases where a researcher is a pioneer in a specialty field, where they may have produced a large volume of work on a "niche" topic that comprises a large fraction of all of the work.
There are certainly cases where a researcher is invited to write a survey or review paper or one that covers the evolution of a specific research topic.
It is possible that these two could be concurrent.
So asking about a single instance without describing the nature of the paper along those two axes leaves open the possibility that this is by design, i.e. what the journals' editor(s) hoped for.
Another paper of his has 96 citations with 87 being self. Is this normal or excessive?
Ya, could be, unless the concurrence happened twice; the author was invited to write two survey or review papers and you happened to find both.
Yes, it is normal and it is the equivalent of doping in sport.
If your competing organisation is self-citing, and both your competitor and you are equally piffle in doing research, then the only way to survive for you is to self-cite as well.
This has two interesting consequences:
- the larger the research group the higher the number of artificial citations the participants can get.
- the doped paper that go through blind peer review is de facto not anonymous (the self citation points eloquently at the authoring research group): two competing organisations can help each others to dope up the number of citations (I approve yours self-citing paper, you approve mine).
The positive note is that this mechanism is only doping up the bottom. Honest and really breakthrough research work have a number of citations that is unreachable for the self-citing research groups.
For example the paper by Nakamoto, "Bitcoin: A Peer-to-Peer Electronic Cash System" has alone more citations than the total number of citations of the self-citing professor in his whole career.