Publication efficiency

high-resolution terminology - matching measurements at high-resolution

Publication efficiency

Description

Publication efficiency is the fraction F_r,a of reproducible publications N_r which are among the number N_a of publications that receive attention and meaningful interpretation, per total count N_p of all published communications. Publication efficiency F_r,a = F_r·F_a is low due to (1) the reproducibility crisis expressed as low reproducibility efficiency F_r = N_r/N_p, and (2) the inflation crisis expressed as low attention efficiency F_a = N_a/N_p. Estimates of these partial efficiencies vary from field to field. With F_r=0.15 and F_a=0.05, the current publication efficiency is as low as 0.0075, or only 0.75 % of all presently published communictions are reproducible and receive attention and meaningful interpretations. Reduction of the number of irreproducible zero-value publications is the most effective measure to reduce the paper mass excess (PME).

Abbreviation: F_r,a

Reference: MitoFit 2020.4

Communicated by Gnaiger Erich (2020-07-14) last update 2020-07-23
in: Anastrophe XX Entity X and elementary unit x of A X-mass Carol

Canon IIIIII: To publish or buglish — a pathway to perish

Our "Age of Anxiety" is, in great part, the result of trying to do today's job with yesterday's tools — with yesterday's concepts. — Marshall McLuhan Herbert, Fiore Quentin (1967) The Medium is the Massage — »Bioblast link«.

Communication and value of science has been based on paper, from the papyrus of ancient Agypt to the transformations that enlarged the pressure to publish and were brought about by the printing press of Johannes Gutenberg. Metal typesetting used fonts of different sizes with a fixed metal mass of each sort of type in a typeface. Today the body mass index of a printed character is called the weight of a font, which is the thickness of the graphical outlines divided by the height of a character. Pressing a button on a photocopy machine produces or rather consumes a lot of papers in the age of the xerographic (dry graphic) technology. In the modern digital world of the internet and pdf files we are still using the 15^th century vocabulary of the printing press when talking about a scientific paper that makes or breaks a scientific career in the service of impact factors, and we are still referring to the physical transport by courier services when spelling out pdf as portable data format. Scientists have to have the paper in press to make an impression on the scientific community. Even preprints give the impression, that printing is performed by application of mechanical pressure. The digital revolution has transformed the culture of scientific publication in physics and mathematics, but a fundamental liberation from a concept fixed on paper(s) is yet to be realized in the biomedical field (Gnaiger 2019 MitoFit Preprint Arch Editorial).

Every scientific communication of value requires realization in the sense of the word that has more than two sides: (1) a physical realization of the communication as a publication available in one or several media (e.g., video and pdf version of a publication), (2) a virtual realization of the publication as the cognitive process of decoding the potential meassage in the publication media into a realized message, received in the mental media of somebody who minds to comprehend the meaning of the message, and (3) back in an Ouroboros loop to a physical realization of the communication in the form of an innovation, as a new artefact with value in the real world — e.g., a new motivation procedure to increase the number of physically active people; a new drug with less negative side effects; a new counting or measuring instrument that is more economical and produces less instrumental artefacts.

Every step in these three phases of realization of a scientific innovation is valuable, and is evaluated differently by individual people in their own way on their way to success. Collectively, however, only real-world artefacts or facts have value in the global fight against the crises of injustice in peace and war, global warming, irreversible loss of natural and cultural diversity, overpopulation of the sick and underpopulation of the healthy. Although these real-world artefacts resulting from the realization of scientific innovation may be positive as the goods of cultural tools or negative as the bads of war within the species and against other species and resources, the scientific innovations realized as artefacts or facts are drivers of human cultural evolution. Therefore, a reflection on the realization potential and validation of a scientific communication has its value.

"No one knows how many scientific journals there are, but several estimates point to around 30,000, with close to two million articles published each year." — Altbach Philip G, de Wit Hans (2018) Too much academic research is being published. Univ World News retrieved 2020-07-14. — This makes close to 5 500 publications per day. If you are working during the week and reading during the weekend, then there are more than 19000 new publications per weekend day.

Compared to browsing through publications in the (physical room of a) library, browsing through the internet and specifically PubMed has changed retrieval of literature fundamentally. Table 6.1 summarizes some examples relevant for the X-mass Carol.

Our official culture is striving to force the new media to do the work of the old. — Marshall McLuhan Herbert, Fiore Quentin (1967) The Medium is the Massage — »Bioblast link«.

Table 6.1. Publication overload: PubMed counts for publications in 2019 and 2009 on BMI and mitochondrial search terms individually (#1 and #2) and combined (#3). There was a 1.7- to 1.8-fold increase within 10 years in all three categories, indicating that growth in the more prominent research field BMI (mt/BMI=0.8) was as high as in the mt-research field. Only 0.43 % of all BMI-linked publications included the term mitochondria.

Search term #	PubMed Search term	Count 2019	Count 2009	Count per day 2019	Count per weekend day 2019
		x	x	x/day	x/weekend day
1	BMI or Body mass index	25310	13861	69	243
2	mitochondr*	19994	11583	55	192
3	1 and 2	108	62	0.3	1.0

Figure 6.1. Number of publications on mitochondria per year. Nobel price laureates are marked who are particularly relevant for mitochondrial physiology and bioenergetics, and the thermodynamics of irreversible processes. Data on publication counts retrieved from PubMed 2020-07-17.

We face a massive problem. The counts in Table 6.1 do not allow us to turn the pages to the list of references and re-read the messages in the cited publications. The probability of finding a (truly) groundbreaking innovation contained in a rejected or published paper is close to but not equal to zero. Therefore, only a high number of published papers realized by a readership of open minds can manage to increase the probability, that the 'true' innovation, which is not only rare but extremeley difficult to comprehend, will be realized or implemented and applied in the real world. Highly simplified models of 'Meaningful-innovation-Publications' (MiP) provide guidelines for reflecting on and optimization of publication strategies.

Paradise lost — the static MiP model

Figure 6.2. New books per million inhabitants per year. Retrieved from https://ourworldindata.org 2020-07-21.

The printing press put an end to the era when books were rare (Figure 6.2), were reproduced as prayers and songs, when canonical repeats in carols aided the reproducibility of the text in memory, and when the pictures in a book were a medium to translate the written word into a message for the illiterate. We live in a paradise of reproducibility of printed materials by simply copying digital files on electronic devices or papers on inexpensive copymachines. Nevertheless, Open Access in science is an unfulfilled dream in the scientific publishing business generating revenue of $25 billion USD with 1.5 million articles published/year (Triggle 2017 Drug Dev Res). Reproducing the publications for a wider and wider audience should increase the number of critical readers and the feedback to authors with suggestions for corrections of any published errors, thus potentially reducing the number of erroneous publications. However, the flood of publications expelled science from the never-existing paradise of scientific reproducibility. The static MiP model analyzes the combined effects of the inflation crisis and reproducibility crisis on publication efficiency with an illusionary view to the paradise or heaven of reproducibility.

1. The number of published papers per unit of time (per year, per day, per weekend day) is N_p. The time unit has to be specified.
2. The number of irreproducible publications N_z of zero positive but potentially negative value is high, with an irreproducibility ratio F_z/p = N_z/N_p ≈ 0.85 (Ioannidis et al 2014).
3. The number of valuable, reproducible publications N_r can be split into innovative N_i and confirmative publications N_c. Bayesian statistics provides a solid proof of concept, that it is irresponsible scientific commercialism, if confirmative publications are suppressed in so-called 'high-impact' journals. The immens value of methodologically sound confirmative communications requires full appreciation, as does the importance of reproducibly contradictory results. It is difficult but important to define the drivers of scientific commercialism among authors, editors, and businesses, and to distinguish re-search oriented publishing of scientific literature from commerce oriented buglishing in low- or high-impact journals which are predatory to the same extent. The underrating of methodologically and statistically rigorous confirmative and contradictory results promotes escape strategies of presenting conformative results, which please the commercialism of publication by mainstream-delusion. These considerations justify an emphasis on reproducible publications N_r — including publications providing the test of reproducibility of published innovative results — and a simple discrimination between reproducible and zero-value publications, N_p = N_r+N_z. The reproducibility efficiency is F_r/p = N_r/N_p.
4. The reproducibility crisis expressed by the low reproducibility efficiency F_r/p has to be put into the context of a hyper-exponentially growing inflation crisis of publication. The count N_p of all publications is split into a count of publications N_a that receive attention and meaningful interpretation, and a count of publications N_n that is neglected, N_p = N_a+N_n. If N_p exceeds N_a, then there is an inflation cisis of producing a paper mass excess (PME), as a process which diminishes the value of each elementary paper U_p. The inflation crisis is expressed by the low attention efficiency F_a/p = N_a/N_p. The concepts of PME and BME are not isomorphic.

The introductory analysis of the reproducibility and inflation crisis (r&i-crisis) of scientific publications is summarized in the following definitions and equations, all of which are oversimplified symbols of reality.

(Eq. 6.1a) Publication count r-crisis: Np = Nr + Nz
(Eq. 6.1b) Reproducible publication count: Nr = Np - Nz

(Eq. 6.2a) Publication count i-crisis: Np = Na + Nn
(Eq. 6.2b) Attention publication count: Na = Np - Nn

(Eq. 6.3a) Reproducibility efficiency: Fr/p = Nr / Np
(Eq. 6.3b) Reproducibility in r-crisis: Nr = Fr/p · Np

(Eq. 6.4a) Attention efficiency: Fa/p = Na / Np
(Eq. 6.4b) Attention in i-crisis: Na = Fa/p · Np

Figure 6.3. Decline of publication efficiency as a function of increasing irreproducible publications N_z* (left to right). Counts N_X* and efficiencies or fractions F* with the asterisc show the results of various levels of improvement from the baseline state with N_z of irreproducible publications and intermediary states with N_z* decreasing to the standard state with N_z° = 0 x (right to left). N_p is the total count of published communications at the current baseline. The reproducibility efficiency is F_r/p=0.15 at the current baseline (red arrow). As the number of irreproducible publications N_z* declines while the number of reproducible publications remains constant at N_r*=N_r, the total number of publications N_p* declines linearly with decreasing N_z* to a minimum fraction N_p*/N_p=F_r/p (dotted arrow). The reproducibility efficiency F_r/p* increases as a hyperbolic function of the declining N_z*. N_z*/N_r* can be understood as the relative inhibitor concentration with a hyperbolic (Michaelis-Menten kinetic) inhibitory effect on the reproducibility efficiency F_r,p*=N_r*/N_p* (Eq. 6.9). The attention efficiency F_a/p* increases as a hyperbolic function of the reduction of N_z* to a maximum F_a/p° = N_a/N_r = 0.33, due to the steady decline of the total count of published communications N_p* as N_z* declines to zero x. The target of optimization is the count N_r,a* or the normalized count F_r,a/p* = N_r,a*/N_p, which contains the quadratic term (N_z/p/N_p)². F_r,a/p* demonstrates the primary importance of cutting down the number or irreproducible publications, which are identified as an inhibitory scientific output and as such not of zero value, but of negative value.

Publications are a currency. Publication metrics is concerned with putting a numerical value on the currency. At a time of currency devaluation, strategies are required to bring the progressive inflation to a halt, whereas the contrary is achieved by elaborating journal impact factors or h-indices which forge a scientific career into a number and into an addicition and craving to increase the number of publications. Even increasing the number N_r of potentially reproducible publications is a lost investment if they get burried under the avalanche of the sheer number of new publications. Advertising for more attention, in turn, is meaningless at a low reproducibility efficiency. Ioannidis et al 2014 suggest that the fraction of irreproducible results published may be as high as 85 %. The corresponding fraction or efficiency F_r/p ≈ 0.15 is even an optimistic estimate in some areas of research. Publication counts such as those in Table 6.1 lead to estimates of an attention efficiency far below the reproducibility efficiency, and a value of 5 % is just taken as an example (F_a/p = 0.05). The r- and i-crises are not only the two sides of the publication coin, but they have a multiplication effect on the diminishing reproducibility-attention efficiency F_r,a and correspondingly low number N_r,a of reproducible publications which actually receive attention and meaningful interpretation. From Eq. 6.4, N_r,a is calculated as:

(Eq. 6.5) Nr,a = Fa/p · Nr

Substiting Eq. 6.3b for N_r,

(Eq. 6.6) Nr,a = Fr/p·Fa/p·Np = Fr/p · Na

shows that N_r,a is diminished by the product F_r/p·F_a/p. This product is the reproducibility-attention efficiency F_r,a,

(Eq. 6.7) Fr,a = Fr/p·Fa/p

F_r,a equals 0.0075 or 0.75 % of the total publication count when inserting the estimates given above — an enormous inefficiency.

Rearranging Eq. 6.6 and inserting Eq. 6.1a for N_p yields:

(Eq. 6.8) Nr,a = Fr/p · [Fa/p·(Nr+Nz)] 

Eq. 6.8 illustrates that a simple mathematical equation can be as ambiguous as simple words when interpreted without attention to context. It looks paradoxical at first sight, as if an increase of zero-value publications N_z could exert a positive effect on the desirable number N_r,a of reproducible publications that receive attention. The mathematical medium 'equation' gives a meaningless message, if not interpreted properly by recognition of recursive elements, as shown in Figure 6.3. N_r,a is strongly controlled by reproducibility efficiency F_r. Therefore, cutting down the count N_z of irreproducible zero-value publications is the top priority.

Consider the paradise of reproducibility when N_z° = 0 x. Then — at a heavenly reproducibility efficiency of one — the publication efficiency is F_a/p° = F_a/p/F_r/p = 0.05/0.15, increasing from 0.0075 to 0.33 or 44 times. How do we get from heaven to hell of irreproducibility? Figure 6.3 shows the trajectories of efficiencies declining with an increase of irresproducible zero-value publications N_z. The variable shown with an asterisc (*) are the trajectories between the standard state at F_r/p°=1 and the current state assuming F_r/p=0.15. The following parameters are fixed in the static model: (1) the number N_r* of reproducible publications is constant at 0.15 times the current total publication count N_p, N_r*/N_p=0.15; and (2) the number of N_a* of publications receiving attention is constant at 0.05 times the current total publication count N_p, N_a*/N_p=0.05. Even at a fraction N_p°/N_p = F_r/p (Figure 6.3, dotted arrow), the attention capacity is insufficient, although 33 % instead of 0.75 % of the reproducible publications N°_p=N°_r receive attention.

1. The reproducibility efficiency declines as a hyperbolic function of N_z*/N_p, where F_r/p=0.15 is the constant parameter:

(Eq. 6.9) Fr/p* = 0.15 / (0.15 + Nz*/Np)

This can be expressed in the form of a familiar hyperbolic inhibition equation,

(Eq. 6.10) Fr/p* = Fr/p(max)* / (Fr/p(max)* + Nz*/Nr)*

where F_r/p(max)*=1 (maximum reproducibility efficiency F_r/p(max)°), and the inhibition constant is K_i(50)=1, which means that F_r/p*=0.5 at N_z*/N_r*=K_i(50)=1, or at N_z*/N_p=F_r/p (Figure 6.3, dashed arrow). Thus N_z*/N_r* is an effective inhibitor concentration causing a steep decline of publication efficiency.

2. The attention efficiency declines as a hyperbolic function of N_z*/N_p, where F_a/p=0.05 and F_r/p=0.15 are the constant parameters:

(Eq. 6.11) Fa/p* = 0.05 / (0.15 + Nz*/Np)

The maximum value of F_a/p* equals F_a/p/F_r/p=0.05/0.15=0.33.

3. The reproducibility-attention efficiency is the target of optimization. F_r,a/p* declines as a quadratic hyperbolic function of N_z*/N_p, where F_a/p·F_r/p=0.05·0.15=0.0075 and F_r/p=0.15 are the constant parameters:

(Eq. 6.12) Fr,a/p* = (0.05·0.15) / (0.15 + Nz*/Np)2

The maximum value of F_r,a/p* equals F_a/p/F_r/p=0.05/0.15=0.33. Increased efficiency is certainly a target at a time of limited resources and publication overload, and diminishing irreproducible publications in Figure 6.3 (from right to left) cuts down this superfluous overload. In the static MiP model, the count N_p* of total publications declines steeply by gradual elimination of irreproducible publications, as shown by the straight line of N_p*/N_p (N_p is a constant). But this drastic cut of publication count provides scope for an increasing count N_r,a of reproducible publications which receive attention. The product F_a/p*·N_p* is invariable (see Eq. 6.6), since the declining N_p* is compensated by a corresponding increase of F_a/p*, as long as the capacity for attention to the literature is the limiting factor. Thus N_r,a* increases linearly with the reproducibility efficiency F_r/p.

Reflections on paradise lost

As of 2020, Open Access should not be an issue, but persistent failure is due to the submissions of manuscripts by the scientists 'sponsoring' with public funding the private for-profit journals, by the libraries 'sponsoring' with public funding the private for-profit journals, and by the academic system supported by public funding exerting enormous pressures on scientists to support the private for-profit journals. The academic system not only supports but is the driver of private for-profit journalism by accounting publications in high-impact journals irrespective of their irreproducibility. Every scientist knows about this conundrum. Every early or not-so-early career investigator is trapped in this conundrum. Most scientists have the privilege of having been trained in institutions heavily supported by public funding. This puts scientist into a responsibility to reflect on the knowlegde received in the training, and to act responsibly. The soldier received training for action in combat and for not asking any questions. Academic training is on asking questions. But when it comes to publishing, academics are trained to publish without asking questions — soldiers in the army of for-profit journals. And since some scientists are clever in business, they run scientific societies with for-profit journals, which mindlessly compete with the private for-profit journals: to support the scientists in their career based on publication counts, and hounour publication counts in competition for the journal's success.

Publication pressure on scientists puts authors with a perspective on their careers into the conflict of setting priorities between their own judgement and the necessity of pleasing reviewers and editors. "The authors declare that there are no conflicts of interest" is pure Humbug in most cases, except if these authors declare, that there are no careers of interest. The scientific community and the public would save time and money and natural resources, if scientists would take action in their social responsibility of reducing publication pollution and act against the melt-down of the overheated publication climate. Climate change requires political action based on the conviction of a democratic majority. Career-scientists, however, are under publication pressure and have no time for democratic lobbying.

The static MiP model illustrates the extent of 'paradise lost', but does not provide realistic guidelines for the future. Publication hell is the black hole where all communications are irreproducible, irrespective of irreproducible results or failure of all copying devices to reproduce a published paper for dissemination. A dynamic MiP model does not lead out of a black hole, but starts at the actual pit of publication efficiencies described by the static model. The dynamic MiP model incorporates interventions addressing the reproducibility-inflation or r&i-crisis. The single r- or i-strategy will not work, as shown already in the static model (Figure 6.3).

Out of hell — a dynamic MiP model

Figure 6.4. Dynamic strategies to increase publication efficiency as a function of decreasing irreproducible publications N_z* (right to left). Counts N_X* and efficiencies or fractions F* with the asterisc show the results of various levels of improvement from the baseline state with N_z of irreproducible publications and intermediary states with N_z* decreasing to the standard state with N_z° = 0 x. N_p is the total count of published communications at the current baseline. The reproducibility efficiency is F_r/p=0.15 at the current baseline (red arrow). As the number of irreproducible publications N_z* declines, the number of reproducible publications increases at half the rate of the decline of N_z*. The total number of publications N_p* declines less steeply than in the static model, whereas the reproducibility efficiency F_r/p* increases steeply in the early phase of the reduction of N_z*, transforming every two inhibitory N_z* into one constructive N_r*. The attention efficiency F_a/p* increases as steeply as F_r,p*, since more confirmative N_r* can be counted, which receive increasing attention and are quickly understood. Compared to the static model, the success of increased publication efficiency F_r,a/p* is immediately realized as N_z* are substituted at 50 % by N_r*, thus enforcing the motivation to get out of the r&i-crisis by fighting even more effectively against N_z*. At N_z*/N_p of 0.15, the target of optimization F_r,a/p* increases from 0.05 to 0.08 in the static model, but to 0.47 in the dynamic model, which is a 5.7-fold higher improvement due to positive feedback.

Paradise lost is not new. But most publication metrics are content with or even re-inforce the established for-profit journalism without provision of guidelines out of the publication efficiency pit caused by the r&i-crisis. The dynamic MiP model illustrates a pattern based on two inevitably necessary but possibly not sufficient traits: The decrease of the fraction of zero-value publications is not merely due to preventing irreproducible results from being published, but 50 % of the decline is driven by converting them to valuable reproducible reports. Thus the total publication count declines less steeply compared to the static model. This enables the attention efficiency F_a/p* to increase as steeply as F_a/p*, since the reproducible reports can be more easily understood in a shorter time, and the motivation to pay attention to publications increases with the quality of the total publication count.

After publication, individual communications should be marked with a reproducibility factor. Proof of irreproducibility is a difficult task, but science is not easy after all. To simplify, a publication as the elementary U_p may be rated as irreproducible. Proof of irreproducibility is a valuable task which needs to be honoured. To do so, any non-predatory journal would have to promote solid proof of irreproducibility by (1) publication of this proof, (2) downgrading the individual irreproducible publication with a visible tag, and (3) implementing an irreproducibility factor for the journal.

There is no theoretical proof of the theorem, that rejected papers contain more or less innovative, reproducible, and valuable statements compared to published papers. Only published papers are publications that can be evaluated by the public. If there is overwhelming (hypothetically true) innovation in a manuscript, the public cannot evaluate its message, because a (true) innovation is usually quite difficult to understand. If a rejected manuscript contains (perhaps) too much innovation and is, therefore, rejected and does not make it to a publication, the public cannot evaluate the non-publication. The reviewers rejecting the manuscript are a small sample of all potential experts, who are prevented from seeing and judging the rejected manuscript. The Open Access preprint communication is the perfect tool available already for several decades, for sharing scientific information independent of anonymous peer-review, and instead aimed at Open Access community-peer review. The static MiP model does not take these opportunities into account. There are countless extensions of the static MiP model, one of which is presented as a very simple example without claim to represent or reproduce the real world.

Workin progress

Supplement

• MiP models in Excel: File:Publication efficiency.xlsx

MitoPedia topics: Preprints, BEC