Science desperately needs disruptive innovation

Disruption – the holy grail of the hi-tech industry – is actively sought by academic institutes, research funds and governments wishing to promote science and monetize its fruits. Nonetheless, there is no methodological framework for disrupting science. Studies seeking this framework focus on Scientific conduct without integrating the vast hi-tech disruption methodologies and experience. To cross this barrier, we developed a new Disruption Index (DI), enabling cross-analyses between hi-tech and science for the first time. The two fields show similar disruption patterns. However, data show that while hi-tech quickly identifies and harnesses innovation advancements, science lags in both and is highly affected by exogenous shifts. Decision-makers can leverage the framework suggested here to assimilate the hi-tech disruption mindset in science.


Introduction
Incremental Progression (InP) is the fundamental scientific methodology pushing humanity forward from one discovery to the next. It guides the perception of scientists of all levels and disciplines to evolve step by step, ensures continuous activity and gradual progress. This systematic mechanism is highly effective within a defined context, or paradigm, when low-hanging fruits are discovered one after the other. Its effectiveness placed the InP in the scientific comfort zone.
However, InP also tunnels scientists to delve into niches where they become experts while narrowing their scope of thinking. Higher-hanging fruits, such as complex phenomena exceeding one's specialty (e.g., climate change multi-effect), became more challenging to study and explain. As a result, some scientific disciplines reached stagnation and scientific research is much less effective (Collison and Nielsen, 2018;Park et al., 2023).
The hi-tech world, especially technological startups and venture-capital funds, understands the high alternative cost of such stagnation -loss occurring by not trying other, less obvious, directions. They, therefore, proactively pursue innovation that will shake existing knowledge and create new realities. This pursuit is expressed by investing in paths that will hopefully lead to Disruptions (i.e., breakthroughs), jumpstarting the entire market, and initiating a swarm of consecutive investments. Notable examples of technological disruptions are the Internet, smartphones, and most recently Generative AI, which affect various aspects of our lives.
Many governments and funding agencies acknowledged scientific stagnation as a strategic threat to economic growth.
They invest in scientific research, despite its plummeting effectiveness (Bloom et al., 2020). These bodies understand the importance of disruptive factors, and thus seek ways to increase innovation (Ip) and invest in high-risk, high-gain projects (e.g., the European Research Council funding schemes, https://erc.europa.eu/). The investments highlight directions believed to produce breakthroughs, or in other words, disrupt science. However, breakthroughs are only proven (or not) in hindsight since disruptive thinking is yet to be structured and understood.
Our research hypothesis claims that while InP is ingrained in the modus operandi of most scientific and industrial paradigms, disruptive thinking, which jumpstarts these fields from one phase to another, remains untaught, serendipitous, and yet highly desirable. Previous studies aiming to define the dynamics of scientific breakthroughs analyzed the effect of research team sizes, the numbers and impact of citations and patents, the depletion of low-hanging fruit and the de-focus of scientific incentives, among other directions (Collison and Nielsen, 2018;Wu et al., 2019;Bhattacharya and Packalen, 2020;Bloom et al., 2020;Bhaskar, 2021;Bornmann et al., 2020a;Bornmann et al., 2020b;Min et al., 2021;Xu et al., 2022;Park et al., 2023;Shepherd, 2023). Since these studies focus on scientific conduct, they do not enable a comparison between hi-tech and science. To cross this barrier, we define a new Disruption Index (DI) and apply it to both fields, while analyzing similarity in disruption patterns (see next section for explanations). The new index and terminology will make disruptive thinking part of the scientific toolbox rather than a serendipitous surprise, thereby helping to accelerate the progress of both science and hi-tech. Qeios,

Results
Data from hi-tech and science show a similar four-stage pattern with some variations: The hi-tech dataset is based on the number of investments in 15 startup categories between 1991-2020 ( Fig. 1 behaviors. While the Dollar amount of investments is an obvious proxy for the activity in each category, it is highly affected by outliers (i.e., substantial investment), inflation, and the inflow rate of funds to the market. On the other hand, the number of investments, regardless of size, can serve as a proxy for market-wide activity, and a tool for identifying disruption. Therefore, the new DI was based on the number of investments normalized by their total number for the analyzed period.
In the first 8-12 years, the hi-tech DI slowly increases (i.e., Exploration stage) until a particular year when values begin to grow exponentially over1-5 years ( Fig. 1; Exploitation stage). This trend change marks a disruption. The growth culminates into a 1-3 year of relatively constant DI values (Plateau stage), followed by a sharp decrease (Exhaustion stage). This four-stage pattern appears in three groups (Fig. 1). Disruption of the light blue group (2009) Table 1).
The science DI represents two fields. DI of the technological-related science (tech-science; light blue in Fig. 2) is calculated based on the number of publications in Quantum Computing and Artificial Intelligence between 1991-2020.
This DI behaves similarly to the hi-tech Exploration and Exploitation stages -it gradually increased between the late 1990s and early 2010s, and sharply steepened in 2017-2018. These young disciplines have not yet reached a Plateau. DI of less the technological disciplines (remote-science; light brown in Fig. 2) is calculated based on the 10,000 most cited peer-reviewed papers between 1984-2020 in Geosciences, Economics, Psychology, Mathematics, and Statistics. This DI citation percentage expresses the assimilation level of research insights (a common measure of academic impact) and the financial investment in research and education (e.g., infrastructure and training of graduate students). The DI shows a four-stage pattern, similar to those described above (Figs. 1, 2), despite the weak relations between its five disciplines.
The citation percentage gradually increased from the 1980s to the mid-1990s (i.e., Exploration), then grew faster over ~5year (i.e., Exploitation). The Plateau lasted between 2000-2010, followed by a DI decline to values lower than the initial ones (i.e., Exhaustion). Correlations between the DI of the five disciplines yield values of 67%-97% (

Discussion
Despite the differences between the various fields examined, the data show their behavior to be similar. The four-stage pattern identified here appears in both hi-tech and science, with variations in the duration and intensity of modifications ( Fig. 3). During the Exploration stage the steady and moderate growth in hi-tech and tech-science DI (Figs. 1, 2) indicates the market cautiously seeks disruption without drastic deviations. Once disruption occurs, both fields quickly identify and evaluate its potential while bursting into action. The shorter Exploration stage and sharper Exploitation ascend in the gray hi-tech group show that disruption identification and evaluation improved over time. However, the more Deep-Tech the discipline is, the faster tech giants exploit its disruption once it occurs, and attract much of the activity. This drastically decreases the number of investments, leading to the Plateau stage. The tech-science gains directly and indirectly from the increasing number of investments and becomes more productive. Based on the four-stage pattern identified here, Artificial Intelligence, Quantum Computing, and other tech-science disciplines are expected to reach their Plateau stage in 2-5 years (Fig. 3). All remote-science disciplines show similar behavior in both the duration of the stages and their intensity, which is peculiar (Fig. 2). Why would the distant disciplines of Geosciences, Economics, Psychology, Mathematics, and Statistics show a synchronized behavior when each undergoes substantial modifications over the years? A possible explanation is that they all are equally more sensitive to exogenous processes than endogenous ones. During the exploration stage, citations are influenced by many intradisciplinary directions and trends, but at a certain point, the citations percentage sharply steepens in all disciplines. Similar to the hi-tech domain, the high citation percentage during exploitation represents a sudden expansion of directions opened by disruptive event or events. The Plateau begins when citations reach a Status Quo, i.e., certain studies are repeatedly cited as the pioneering of the discipline. This behavior is equivalent to the takeover by tech giants. The newer cited studies form finer variation superimposed on the Plateau (Fig.   2). When the pioneering studies become less relevant, the Plateau ends and the Exhaustion stage begins while the disciplines await new Internal breakthroughs. Nonetheless, the highly synchronized behavior of the distant disciplines suggests they represent the state of science in general. While universities and funding sources acknowledge that science has reached stagnation and encourages potential breakthroughs from within the discipline, our data suggest that the eagerly awaited disruption will most likely arrive from exogenous events.
The main driver for this systemic behavior is probably technology. In the late 1990s, the Internet became a major force that changed research activity. Scientists moved from citing printed papers (from libraries, photocopies, and pre-prints sent by snail mail) to searching them online. This may have shifted the remote-science disciplines from Exploration to Exploitation (Figs. 2, 3). However, after about five years, science reached another plateau, and it could be argued that accessibility promoted by the Internet only shifted science from one plateau to another while it remained stagnant.
The end of the Plateau (2008-2010) may represent the immersion stage in which publications build their influence. Their delayed influence further stresses how InP inhibits the citation of disruptive papers. However, the Exhaustion stage began almost simultaneously in all disciplines, pointing at exogenous circumstances. We attribute this shift to improvements in computing power which allowed better data acquisition and stronger analytical tools able to crunch considerably larger datasets. These exogenous technological advancements made the frequently cited pioneering studies obsolete. Science entered a new exploration stage where the analytical tools created new low-hanging fruits to be harvested. Scientists rely more on data and quantitative models, and less on established theories and conceptual ideas. This 10-15 years trend questions the effectiveness of the citation system widely used for evaluating studies for funding and promotions. Today, the sweeping introduction of Artificial Intelligence and Machine Learning tools will likely create another systemic change across all scientific disciplines. Again -an exogenous influence.
4.1. Why did the prevailing InP methodology lose its effectiveness?
The InP methodology is most effective during the Exploration and Plateau stages, when basic knowledge is being built.
For example, when mapping and describing terra incognita in any discipline and understanding the relations between its dominant processes. InP is effective in these cases since knowledge maps are still underdeveloped, and there are many low-hanging fruits to discover. Nevertheless, when the low-hanging fruits become scarcer, and the discipline moves to an Exhaustion stage, InP becomes less productive.
In InP, processes transforming status A to B are described using catalysts and inhibitors. Such Causal Logic, coupling a cause with an outcome, underlies most studies and the training of future scientists. For example, the influence of a compound on a biological process or a chain reaction induced by an underwater earthquake. However, as scientists specialize in their discipline (i.e., dig deeper into an InP process), they realize that Causal Logic can only provide a partial explanation, and that deduction by elimination does not necessarily lead to an acceptable solution. To bridge over the gap in understanding, studies simplify the cases by neglecting components or reducing dimensions. These simplifications fit the InP specialization, not the phenomenon under investigation. We suggest that most case studies examined in science are derived from a larger multi-complement, multi-dimensional complex system of interweaving processes that dynamically affect each other. However, the InP and traditional Causal Logic only work in "Silos" (i.e., confined to the InP universe) and cannot deal with several components simultaneously.
Here InP promotes a non-disruptive and less-exploratory science, which relies on the availability of low-hanging fruits (Fig. 3).
Its ineffectiveness leads to one of the greatest crises in all scientific disciplines today. Institutes and funding agencies that understand the crisis and encourage cross-and multi-disciplinary studies on the individual scientist level (yet, these scientists are products of the InP specialization process). They also began to cautiously question the fixated and archaic disciplinary division (yet, at the same time, reject projects that jump too many steps ahead). Walls between disciplinary teaching programs begin to (slowly) break. This positive approach is insufficient to prevent science from slowing down, since most scientists still follow the InP methodology, which hinders disruptive conceptual leaps.
On top of that, although universities primarily promote scientific findings, they await to monetize innovations and inventions. Therefore, they implement a supportive innovation strategy to drive research into business-related areas.
While this approach may catalyze economic gains, it merely invests in gradually improving existing research that follows The progress of science can therefore be described as an interplay between Disruptive scientific leaps on top of an ongoing InP (Fig. 3). InP bursts science into Exploitation when low-hanging fruits are abundant, but when they become scarce InP reacts with Exhaustion. InP does not form disruption itself. Still, only InP efforts have developed into a wellestablished research process that includes funding resources, publication infrastructure and collaboration frameworks. On the other corner, Disruptive research remains anecdotal and is only beginning to be recognized as a crucial force that pulls science forward. The dizzying pace at which the high-tech industry is advancing and creating Disruption is unmatched by universities and research institutions. Despite their efforts to promote innovation and entrepreneurship, universities do not produce considerable breakthroughs. Their main discoveries are made in relatively new areas where little research has been done so far, or data has not been collected in large quantities to date.
Substantial breakthroughs require the challenging of traditional conventions. In hi-tech, this disruptive approach creates breakthroughs, e.g., a new market category that profoundly shifts the daily routine to a new and unpredictable path. The Internet also exemplifies a phase transition (Fig. 1). Its intervention was not merely a technology as described in its early days. The Internet diverted the progress of an immense amount of systems and scientific disciplines to a path that was impossible to predict with pre-Internet knowledge. Science is cautious and its progress relies on knowledge that gradually builds. However, as extensive as pre-transition knowledge may be, it cannot reliably predict post-transition behavior since current "siloed" logic will likely break down, and new apparent causality will emerge.
Today there is no established methodology for creating Disruption in hi-tech (this could be an oxymoron). Nevertheless, the hi-tech industry developed a practical and powerful toolbox that only partially entered universities, mainly through the monetization of research. Decision makers in universities, funding agencies, and research institutes should adopt a venture capital approach to boosting new disciplines, and allowing research pivots (i.e., unexpected turns), based on the understanding of disruption activity and patterns, as manifested in the DI and DP measures. They need to encourage endogenous disruption by promoting a "startup mindset"; endorse exogenous disruption; support non-InP exploration directions; improve the identification of exploration-to-exploitation shifts and rapidly respond to disruption points as hi-tech does; implement a more holistic, Complex-Systems approach to research; improve the incentive to go beyond InP into Disruption; and reduce the heavy reliance on citations as indicators of research disruption potential. All these will help to accelerate scientific disruption, alongside the traditional InP and Causal Logic.