FormalPara Definition

Technological change is seen to be a force of increasing salience for value creation and competitive success of firms. Cycles of emergence of new technology, growth in performance and effectiveness and eventual stagnation or stasis are thought to be recurring phenomena in industries past and present.

Performance as a Predictable Trajectory

Technological performance has been thought of as a monotonically increasing function of effort expended on development and is assumed to follow a rising double exponential or ‘S’-shaped curve. The assumption of slow, rapid and then again slow improvements forms the conceptual cornerstone of many works on technology strategy and innovation.

The shape of performance curves is usually explained by suggesting that in the early days of a technology there is a plethora of possible approaches and directions for development and great uncertainty about which to pursue. Progress will be slow at first due to many failed experiments and dead ends. As early failures and successes build expertise, uncertainties will be resolved and progress accelerated along a more and more clearly defined trajectory. Soon, however, many of the most promising avenues will have been exploited, and physical limits and binding constraints may appear. These may attenuate or even tightly limit further improvements in performance. Much early work attempted to predict technological progress using various techniques for trend extrapolation and for defining limits (Ayres 1969: ch. 6). The problem of multiple performance measures, especially for sophisticated products and systems, was addressed by the creation of various indices and weighting schemes (for example, Saviotti et al. 1982). The reality that performance is rarely depicted by a clear set of points, but rather by a broad smear of data including leading and lagging examples was noted. In an extreme case the differences between leading performance and technology in general use spanned four orders of magnitude (Hilbrink 1989)!

Hirooka (2006: 336–340) usefully relates research trajectories to development trajectories and diffusion trajectories for a number of technologies including solar cells, fuel cells and superconductors. Porter and colleagues (1991: 170–172) summarize the popular idea that an envelope of performance over time can be constructed by stacking S-curves of successive individual technologies, such as the vacuum tube, transistor and integrated circuit, and suggests the resulting envelope will also have a double exponential shape. An early test of this hypothesis led to disturbing results. Far from being predictable the envelope for generations of typesetting technologies was relatively stagnant from 1500 to 1960, but progress in typesetting speeds has been nearly exponential since then with no sign of diminishing (Mohn 1972). Moreover, Mohn (1972: 227) provides data for hot metal linotype setting speeds that negates all of the ideas just summarized. From 1886 speeds increase rapidly until they reach a long plateau, continuing until about 1960, of roughly 10,000 characters per hour. Then, another surge of improvement is observed, just the opposite of expectations regarding limits. This is but the earliest of many contradictions of popular opinion to be covered in more detail below.

As mentioned above, technological performance has been thought of as a monotonically increasing function of effort expended on development. Cumulative effort is difficult to define and to measure, especially in sectors with numerous and often changing competitors. As a consequence, much of the research on technology cycles resorts to plotting performance as a function of cumulative production volume or even simply over time assuming that production quantities and development efforts increase in constant proportions with time (Fusfeld 1970: 308; Sahal 1981: 186–187).

The Importance of Technology Cycles for Strategy

The critical importance for strategy of understanding technology cycles is that most firms invest far too heavily in development efforts long past the time that rewarding improvements might be expected. Further, once threatened, firms typically redouble their investment in familiar concepts and fail to make a transition to an emerging concept. Most research on technology cycles presumes that a technology with superior performance will be preferred by the market. But the potency of a newly introduced technology may be that it offers different performance, enabling a great expansion of the market and openings for new rivals (Cooper and Schendel 1976; Foster 1986).

In his influential book Innovation: the Attacker’s Advantage, Richard Foster gives two examples, both homogeneous products from the chemical and synthetic fibre industries, of performance related to R&D effort. Foster (1986: 157) observes that firms almost inevitably wait too long to attempt a transition from a dominant established technology to an emerging challenger, and that they overestimate their ability to anticipate discontinuities, to identify potential new competitors and to time the onset of a new technology cycle. Building on the widely accepted ideas about the dynamics of performance improvement summarized above he argues firms should switch their investments to new concepts much earlier than they usually do. In essence he believes that the myopia of long-established firms gives attackers a significant advantage during transitions between generations of technologies. Foster cites a long list of products in the container and packaging business whose market positions have been overturned by innovative competitors: glass bottles by steel cans; steel cans by aluminum cans; glass bottles by plastic bottles; plastic-coated milk cartons by plastic jugs; and so forth. In each case, he notes that market leadership passed from one set of firms to another. Today’s leaders, in these cases, were never leaders in the next product generation. ‘I don’t know of any comprehensive statistics that would stand up to academic scrutiny, but my feeling is that leadership changes hands in about seven out of ten cases when discontinuities strike. A change in technology may not be the number-one corporate killer, but it certainly is among the leading causes of corporate ill health’ (Foster 1986: 116). He argues that an optimal pattern of development spending would tend to follow a normal distribution, with the majority of spending concentrated in the centre of the cycle, rather than constantly rising and being concentrated toward the end of product life.

Cooper and Schendel (1976), examining 22 cases of technological discontinuities, observe that threatened firms typically redouble their investment in familiar concepts and fail to make a transition. Often they ride out the failure of the old while making only defensive investments in the new. Cooper and Schendel conclude that such a dual strategy is simply not viable. Their explanation for the observed syndrome is that, ‘decisions about allocating resources to old and new technologies within the organization are loaded with implications for the decision makers; not only are old product lines threatened, but also old skills and positions of influence’ (Cooper and Schendel 1976: 68–69). It is not uncommon for a threatened firm to develop a hybrid or intermediate form of product combining aspects of both new and old. An example was Lockheed’s Electra aircraft, a hybrid of jet engine and propeller. The Electra was not a successful competitor to rapidly evolving turbojets (Girifalco 1991: 112–113). Hybrid strategies seem to have a general history of failure. Examples of failed hybrids are legion from the steaming sailing ship to Thomas Edison’s GEM lamp and to Good-year’s bias-ply belted tire (Utterback 1994).

Irvine and Martin (1984) extend expectations regarding predictable trajectories to the arena of public policy, developing the concept of macro-strategic research. Their work has led to the further development of research milestones or strategic technology road mapping at a national level. If ideas about the predictability of technology cycles and trajectories are incorrect, then such rigorous attempts at planning will lead to errors and misallocations of resources both for corporations, and more importantly for industries and whole economies. The weight of research reviewed below suggests that the most popular ideas about mapping and predicting technology cycles and performance are seriously over simplified and misleading. It is more vital than ever for strategists to understand the changing texture of technology, but in a richer and more nuanced way.

A Critique of Widely Accepted Theory and Applications

Hints of trouble ahead are evident in many of the sources reviewed. These include technologies with linearly increasing performance as well as ones with continuing exponential improvement despite repeated predictions of limits, Moore’s Law being a famous example (Mollick 2006). Moreover, many technologies, which seemingly have reached limits, are dramatically reborn when faced with competition from something new. The gas lighting industry when faced with Edison’s carbon filament incandescent lamp responded by developing an incandescent ‘mantle’ of ceramic filaments improving the efficiency of gas lamps by nearly threefold. ‘For a number of years the potential superiority of the incandescent lamp remained in doubt, and even its survival was sometimes questioned’ (Bright 1949: 127). Edison himself faced with competition from European innovations in metal filaments created a new form of carbon, the ‘GEM’ lamp, which performed nearly as well. Bright (1949: 181) reports that, ‘the filament which resulted in 1904 from Whitney’s work was the greatest improvement made in the carbon lamp since 1884’. These and myriad other examples (in Utterback and Kim 1986; Utterback 1994) are stark evidence not only of firms’ tendency toward retrenchment and tradition when threatened, but also of complacent incremental improvement at other times.

In a searching review of the technology cycle literature including a further detailed study of 14 technologies, Sood and Tellis (2005: 152) conclude that, ‘the results contradict the prediction of a single S-curve. Instead, technological evolution seems to follow a step function, with sharp improvements in performance following long periods of no improvement. Moreover, paths of rival technologies may cross more than once or not at all.’

Koh and Magee (2006) take a more broadly functional approach to technology cycles by focusing on three broad categories; matter, energy and information, and then classify technologies according to the manner in which they operate in different domains such as: transform, transport, store, exchange and control. Their approach provides a way to deal with the varied multidimensionality of individual technologies. In applying their technique to information storage, calculation and communication using a 100-year data series Koh and Magee find generally continuous progress for each functional category independent of the specific underlying technologies dominating at different times. In essence they suggest that the envelope of performance for a succession of technologies in each category is more stable and continuous than is that for any individual cycle. In a later study (2008) of storage, transportation and transformation of energy, Koh and Magee report substantial variability of progress rates is found within given functional categories for energy compared to relatively small variation within any one category for information technology.

In a similar general argument, McNerney and colleagues (2012) make the prediction that the rate of improvement of a technology depends on its design complexity, that is the number of components incorporated and the number of connections among them. The possibility that design complexity may be reduced through time will, they suggest, be correlated with rates of performance improvement. Thus, products having many interconnected parts may advance more rapidly than simpler, more homogeneous, and more integrated examples, and as designs become simpler over time the rate of advance would be expected to slow. Their model and hypotheses are resonant with histories presented in Utterback (1994) in which patterns of innovation for complex products such as electric lighting, typewriters, calculators and computers are contrasted with homogeneous products such as sheet glass and rayon, and products and processes do indeed become simpler and more integrated over time.

Transitions Between Cycles

Much research on technology cycles implies that growth of a new technology comes at the expense of displacing prior art. A key point for this review though is that an innovation often both enables growth into a broader market and prospers through it, rather than displacing prior art in an existing market. As we have seen, a presumption is that technology with superior performance on a traditional figure of merit will be preferred by the market. But the salient feature of a newly introduced technology may be that it offers dimensions of performance not possessed by the traditional offering, allowing the market to be re-framed (Christensen 1992a, b; Levinthal 1998; Kaplan and Tripsas 2008). Thus, radio was first defined as ‘wireless telegraphy’ and for years occupied just a niche market, communication over water, where wired connections were absent. Only when its ability to reach a mass audience with news and music was recognized did its transforming potential become realized. There were no extant competitors with broad reach, and the growth of ‘radio broadcasting’ was explosive.

A widely held belief is that at the time an invading technology first appears an established technology generally offers better performance or cost than does the challenger. The new technology may be viewed objectively as crude, leading to the belief that is will find only limited application. The performance superiority of an established technology may prevail for quite some time, but if the new has real merit, it typically enters a period of rapid improvement – just as the established technology enters a stage of slow improvement. Eventually, the newcomer improves its performance characteristics to the point where they match those of the established technology and rockets past it, still in the midst of rapid improvement.

Utterback (1994: ch. 9) in a meta-analysis of 46 transitions in technology cycles finds that a discontinuous change may drastically increase the aggregate demand for the products of an industry. The replacement of the vacuum tube by the transistor, and later the integrated circuit, has increased the sales of the electronics industry from several billions of dollars to hundreds of billions. The replacement of piston aircraft engines by turbojets has correspondingly dramatically reduced the costs and increased the seat miles flown by commercial aviation. Innovations that broaden the market may create room for new firms to start. Innovation-inspired substitutions conversely may cause established firms to hang on all the more tenaciously, making it extremely difficult for an outsiders to gain a foothold and the cash flow needed to expand and become a player in the industry. Innovations that substitute for established products and processes thus may arise more often inside an industry. Some innovations create a wholly new market niche, encouraging the entry of many new entrants. Here, established firms are unlikely to enter successfully and new firms have greater survival odds.

Christensen (1997), in a comprehensive study of the Winchester disc drive industry, finds that each transition in the leadership of the industry was led by new firms addressing an un-served market segment with simpler and less expensive architectural innovations, rather than with wholly new technology. He observes that performance curves of product variants need not intersect for change in leadership to follow. Firms having different products address a spread of performance levels and demands. While older firms focus on demanding major customers, demand is more elastic and rapidly growing in the un-served market for simpler variants. Demand for performance is better described by a broad rising band rather than a line, just as is performance itself.

In a study of the evolution of laser printers de Figueiredo and Kyle (2006: 242) observe two innovation frontiers; a top frontier that is the traditional ‘make it better, faster’ one; and a bottom frontier of ‘make it cheaper and accessible’. Advances evidently occur at both boundaries, allowing penetration of a broad range of new market segments. They suggest this to be true of many other technologies including integrated circuits, microprocessors and digital cameras.

Technology Cycles as a Function of Experiment and Synthesis of Diverse Inputs

Abernathy and colleagues (1982) suggest that performance, rather than being a monotonically increasing function of effort expended on development, is a function of frequency and diversity of experimentation in the market, and of conditions that encourage experimentation. Within the rich mixture of experimentation and competition at the start of a cycle some centre of gravity eventually forms, usually in the shape of product standards or production practices, that is, a ‘dominant design’. The bases of competition change radically, and firms are put to tests that very few pass. Before too long, the ecology of competing firms changes from many competitors to few. One might consider each firm’s investments and product introductions as experiments, which provide corrective and stimulating feedback to that firm and to the industry about product and market requirements. Thus, the earliest period in the development of a product line or industry, in which few firms participate, would necessarily be a period of relatively slow technical progress and productivity advance. As larger numbers of firms enter the arena, thus broadening the range of experimentation and the definition of the product technology, greater innovation with correspondingly greater technology progress and productivity advance should be expected. Finally, as a few firms come to dominate the industry with superior product technology and productivity, both experimentation and progress would be expected to slow (Utterback 1994).

This is not to say that the narrowing of search never reflects a genuine lack of technological opportunities. But it is to say that, more often than not, firms become so structured that they only search very narrowly. Indeed, it is this structuring that critically limits progress: the more specialized firms are, the more narrowly they tend to search for new opportunities (Abernathy et al. 1982: 7).

Radical innovation, and the growth of new industries, is probably more likely when a firm occupies the confluence or convergence of distinct streams of emerging technology. Research progress at the intersection of fields is more likely to occur when cross-disciplinary new product development teams are designed by the organization and when routines and processes are designed to support cross-disciplinary learning. A confluence of technologies is characterized both by the bringing together of formerly disparate fields of knowledge and by the creation of new product markets. When a confluence of technology streams occurs, rich opportunities for experiment and progress may result (Utterback 1994). These may ultimately lead to an emerging industry and newly dominant firms (Maine et al. 2012).

In summary, amid rapidly changing markets, extended supply chains, and widening sources of competition, dislocations and opportunities frequently arise from surprising sources. We might more creatively think of innovation and firm formation as a process of experimentation in the market. Rather than seeking to reduce uncertainty and to optimize, perhaps we should seek to increase possibilities for experimentation and for broader search and synthesis. Technology cycles then are anything but consistent and predicable trajectories.

See Also