Very interesting questiones, additional question should be :

what possibilities are implemented in these metrics, that allow the real authors

to evaluate faked plagiaristic publications upgraded by the real authors

datasets with only minimal modifications - just in an other laguage. ???

and no real new ideas !!!!!

Current metrics based on other papers citing the paper are almost useless and only have a thin veneer of respectability and reliability. Current metrics are designed to sell more journal articles from the same stable, whether the citations be positive, neutral or negative. As you have identified, the real test is the test of time. Is the work still positively cited next decade? Next century? Why is it so true that many artists, composers and writers (and researchers) are only recognized for their worth and contribution to human knowledge after their deaths? Was the research the start of a revolution in our thinking? Did it add to humankind's lot? Was it excellent, pioneering, groundbreaking research?

In the short term, perhaps the referees' comments could always be appended to each paper reviewed.

Ian Kennedy , Thank you for your reply. In the end, the researcher must decide if they want to strike that balance between chasing metrics (for daily bread) and contributing to the benefit of humanity (for the psychological value in altruism).

Metrics used in online environment can never determine the utility of work/ research. At present knowledge can't determine the status after twenty years as knowledge is accumulating at very fast speed .

Any published/ or written material may be used after decades at that time importance of previous work is understood

Yes, I see this problem.

There are authors who are talented for poetry, and publish their ideas under different auspices several times. There is a need to publish many papers, and if the experimental results are poor, and they are fluent writers - this is the result.

In my case, at a non-university public investigation and research institution, the heads are managers/ merchants, and they are not interested in scientific papers, because there is no income. Thus, I need not publish, and each paper relies on many new data.

The same is the case for contributions to symposia. I have known professors who have been invited as keynote speakers, and presented about the same lecture throughout many years (titles changed) throughout many years - the audience changes anyway. This makes the number of their lectures quite large.

This is a really good question, and almost impossible to be answer with a magic formula, just as the metric tide desire to present itself. The way to answer it is to develop multidimensional studies, which can mesure temporalities, spacialities, time of carreer, networks of relations not only in academia, but in other spheres, among others variables, specially on the whole scientific ecosystem. And of course, change the concern about impact to worry about the circulation of science. That how I am trying to work on.

Excellent question.

And for the answer, it depends on what is being considered as the "real importance of scientific work." Metrics used in online environments may indicate:

- current topics and subjects;

- popular scientific works that have reached high audience;

- public interest;

- circulation and interaction around papers.

- others.

In all cases, in order not to be reduced to another number that cannot have an end in itself, it is necessary to understand the context of these indicators and try to understand who shares the papers; with whom they share; when they are shared; and how they are shared.

It can also be that a paper appears to early, and nobody is interested in it, particularly if this is written in a more or less unusual language. After 20 years or more, there is a hype - and things are invented again, and these papers are very much referenced.

My students have looked for references only 10 years back, and found mainly Chinese papers in English. But it may happen that similar works have been issued 50 years ago in German, Russian, Polish, Italian or else language, and nobody has cited it, but I found something in my personal library about, which Google does not know.

Another problem is the problem of co-authors. Long time ago, I was in a jury to judge the scientific qualification of professors, to whom students had applied for a grant. Criteria were the number of papers in renowned journals. A candidate had 35 scientific workers, and he had more than 50 papers, being maybe just the 7th dwarf of the story. Maybe he would not even have the time to read the final texts, but as a courtesy to the big manager, and to avoid objections, he might have been invited.

Nice question. As several other respondents have noted, we need to be careful not to "throw the baby out with the bathwater". Citation-based metrics are not meaningless, but they do need to be interpreted with caution since they are sensitive to both random and systematic biases, including those caused by the 'citation culture' that the use of these metrics encourages (see Todd, P.A. and Ladle, R.J., 2008. Hidden dangers of a ‘citation culture’. Ethics in Science and Environmental Politics, 8, 13-16).

Returning to your original question, I'm not sure we can read too much into historic patterns since science has changed so much, even within my career (I got my PhD in 1993). Science is so much more collaborative, interdisciplinary and international that, for better or for worse, the days of lone scientists publishing 1 article every 5 years have disappeared. For anyone interested, there is a growing literature on alternative metrics that seek to 'capture' the social/collaborative nature of research - indeed, ResearchGate's reputation metrics show a lot of promise in this respect (e.g. Nicholas, D., Clark, D. and Herman, E., 2016. ResearchGate: reputation uncovered. Learned Publishing, 29(3), pp.173-182; Yu, M.C., Wu, Y.C.J., Alhalabi, W., Kao, H.Y. and Wu, W.H., 2016. ResearchGate: An effective altmetric indicator for active researchers?. Computers in human behavior, 55, pp.1001-1006.).