For veterans of the information technology (IT) industry, the late 1990s was a remarkable time — often referred to as the "Internet revolution" amidst a "New Economy". It was an era when seemingly limitless venture capital (VC) flowed into the coffers of e-commerce firms, some of which had business models that even their ponytailed CEOs could barely explain. VC capital was burned at a phenomenal rate, with more focus on penetrating markets and acquiring "eyeballs", and less on actually making a profit. But within a few short years, the "dot-com bubble" met the sharp pin of reality, and overvalued NASDAQ stocks tumbled as Wall Street lost patience with companies spending cash with little more to show for it than snazzy websites and not-too-snazzy balance sheets.
Since most of the darling firms were Internet-based, by necessity a large portion of the money inflows ended up in IT budgets, as well as IT salaries, though to a lesser extent in the opinion of many now-disenchanted programmers. The general consensus among high-level managers of IT projects, is that their big spending at the time did not pay off in nearly as many successful projects as they had anticipated — or at least, as many as had been promised by technology vendors. Even worse, the tremendous increase in investment, on a broad scale, made little impact in improving upon the high failure rates of IT projects established in earlier decades.
Now that the Internet party has dramatically wound down, the former highflying managers are finding themselves with increasingly circumscribed budgets, while the engineering staff — those still employed in the U.S. — are being asked to do even more with less. In turn, they are likely less willing to sacrifice their evenings and weekends for the company, now that the false promises of stock options are being seen for what they truly are worth, namely, "bird cage lining". These industry participants, now older and wiser, may still wonder at times whether the computer field will ever regain its past glory, blistering growth rates, and envious societal status.
According to author Erik Keller, it's not going to happen. As an IT industry veteran and technology consultant, Keller captures these views in his book Technology Paradise Lost: Why Companies Will Spend Less to Get More from Information Technology, published on 1 April 2004 under the ISBN 978-1932394139 by Manning Publications, whose user group representative kindly provided me with a copy of the book for review. The dust cover blurb may well summarize his position: "American corporations let IT grow until it reached one half of all corporate capital spending by the year 2000. Now, chastened by their spending failures, IT managers are converging on a new consensus: to exploit IT competitively they must use their smarts over big money[…] Counterintuitively, companies that spend less in order to get more from information technology will likely be the big winners." That's quite a claim, and I will examine how well Keller supports it in his book.
The preface of Technology Paradise Lost helps to establish Keller's credentials for covering these topics, as he briefly discusses his background in technical journalism and technology analysis for Gartner — almost a decade in the former field, and over a decade in the latter. It is that second stage of his career that put him in a position to provide advice to over 1000 companies utilizing technology, many of whom ranked in the Fortune 500. He notes that, "When the technology was used well, the results were amazing. There was always a risk of failure, however, when companies jumped into a technology too quickly […] Even when projects were successful in the beginning, they were often followed by unanticipated difficulties downstream" (pages ix-x).
As that exuberant decade culminated in frantic efforts to avoid Y2K catastrophe, Keller was seeing more of those unanticipated difficulties, and "[…]was spending more time helping clients clean up problems than launch new technology initiatives" (page x). In the years that followed, he concluded that a new perspective on IT spending was greatly needed. Specifically, he became quite interested in the questions more CIOs were being forced to ponder: What is IT's appropriate role? What is the right level of IT investment? What benefits should be expected from it?
Keller argues that technology, like the previous four major revolutions, is now moving through an inflection point. Unfortunately, he does not explain to the reader what meaning of "inflection point" he is using. Like so many other terms used by management types, "inflection point" has countless meanings in use nowadays. A common interpretation of the term is that it refers to a defining event that signifies a major change in direction, such as the fall of the Berlin Wall being an inflection point for the unity of the Soviet Union. Using that meaning, what event does Keller have in mind? The reversal of Internet stocks? Even that was a multi-year process.
Again the reader encounters Keller's sizable claim: "companies can move ahead over the next few years without large increases in their IT budgets. The only thing a company needs is a different perspective." (page xii). That prescription sounds suspiciously similar to the oversimplistic advice found in positive thinking self-help books. Keller does not yet make explicit what the different perspective will do for business. Perhaps it should be taken at face value, in that it will allow companies to move ahead without increasing their IT budgets. But is continued progress without budget increases such a massive gain? More significantly, how does that address the larger issues of failed IT projects, to which he alludes earlier? In my opinion, that issue is of much greater consequence.
The book's first chapter, entitled "Paradise Lost?" , begins with the claim that "Massive change occurs quietly, insidiously, and without apparent direction." (page 1). It is not clear how anyone could believe that massive change does not have apparent direction. Directionless change, especially on a large-scale, is oxymoronic. Statements such as this one may appear on the surface to be profound, and yet they crumble under careful consideration. Such products of fuzzy thinking are, unfortunately, typical of American business writings, as well as the PowerPoint presentations of business consultants. One can only hope that this beginning sentence does not reflect upon the care taken in all the subsequent sentences that comprise the book.
The author's brief account of an Antarctic ice shelf collapsing, as a result of the accumulation of many tiny changes, is meant to illustrate a similar deterioration in American IT, in the form of "missed revenue targets, corporate layoffs, and an overall industry malaise" (page 1). But the analogy does not hold up (no pun intended), because the ice shelf collapse was a sharp and relatively quick transition from structural integrity to complete dissolution, preceded only by changes dwarfed by the ice shelf's enormous size. This phenomenal event, which took about five weeks, was truly an inflection point, demarcating the centuries of the ice shelf's steady presence, from what will probably be centuries more of its nonexistence. In glaring contrast, the process of IT spending coming back down to earth, is much more a multi-year reversion to the mean. To put it in engineering terms, the ice shelf's destruction was more like a step function, dropping from 1 to 0 in little time, while IT spending (as a percentage of total corporate expenditures) has roughly traced a bell curve, despite a chart later in the book that looks angular due to so few data points.
The idea that IT spending relative to total budget could ever again approach the level seen during the dot-com mania, appears to me to be either nonsense or unsubstantiated wishful thinking on the part of IT managers and engineers. In fact, prior to reading Keller's book, I had not encountered the thought expressed in any writing, printed or online. Yet, in his opinion, the consensus among most U.S. business managers is that IT (over)spending will resume its upward march at some point. Like unhappy revelers suffering a post-party hangover, some IT veterans may still hold out hope for a return to the days of champagne and stock options. But they must face the cold fact that the excessive IT spending of that era was merely a one-time event, a hope-induced aberration — probably never to be seen again, at least in our lifetimes.
Keller correctly points to some of the reasons why the heady e-commerce binges are not about to return: increasing scrutiny of IT budgets, greater demand for return on investment (ROI), cheaper and simpler solutions, offshoring of software development, lower wages to American programmers, abandonment of failing projects, Internet-based architecture, and adoption of open source software (OSS), such as Linux. Addressing these changes at a more strategic level, Keller notes that, "After years of questionable returns, cost overruns, and increased complexity, companies are pushing financial rigor to IT groups." (page 6).
The first chapter continues with a brief discussion of the general causes of IT project failures, and the new focus on approving only those projects that can quickly show a significant return for the cost and effort. Keller avoids making the common mistake of assuming that most IT expenses are for new systems — whether off-the-shelf, or developed in-house or by outside experts. He cites a statistic that about 70 percent of IT dollars are spent on existing systems, and that most of these legacy applications are a mess, thanks partly to "inconsistent strategic initiatives" (page 9), i.e., management screw-ups. Veteran engineers may find themselves gagging at the overuse of such business-speak terms as "align", but the hard lessons learned are not sugarcoated.
Chapter 1 ends with mention of Sir Ernest Shackleton's remarkable 1914 expedition to Antarctica, to attempt a transcontinental trek. Keller does not explain the relevant point of relating Shackleton's adventure. But, in my view, any book that has at least one of Frank Hurley's incomparable pictures of the Endurance, can't be all bad. Finally, Keller wraps up the chapter by predicting that the next phase in IT will be an evolutionary and pragmatic transition to reduced and more efficient spending.
In his second chapter, Keller outlines the various stages of IT capital investment, from 1960 to 2002, thus encompassing the shifts due to office automation using mainframes, minicomputers, and finally PCs, followed by enterprise resource planning (ERP), Y2K fixes, and the Internet boom. He makes a good point that there is growing overlap of corporate systems, because their developers, managers, and users are resistant to discontinuing any of them that work, even if that means that IT resources have to be used for maintaining these aging and redundant solutions. These additional burdens were not alleviated by the business process reengineering (BPR) movement during the 1990s.
Keller argues that software is not flexible, that business rules captured in software include interrelationships that cannot be changed, and that even small changes can cause system failures. Being a consultant and not a programmer, he is likely unfamiliar with the capabilities of object-oriented languages and methodologies, and the decoupling of systems through the judicious use of application programming interfaces (APIs). The rest of the chapter covers other major causes of IT overspending during the 1990s, including blind implementation of ERP software, increased purchasing of other software applications, Y2K fixes, websites and Internet integration. He then documents how corporate software shifted from being mostly developed internally, to being purchased in the form of packages developed and/or sold by outside vendors. The chapter concludes with short profiles of three different types of software buyers. This book's material does little to enhance the reader's understanding of what needs to be done to achieve the stated goal of greater IT capability at a lower cost; in fact, the entire chapter could be dropped, without ill effect.
In a chapter entitled "Less Bang for the IT Buck", the reader learns of more examples of major corporations making equally major mistakes purchasing software packages touted by vendors and consulting firms, without first verifying the suitability of the software or checking references from existing clients. While the errors covered in the previous chapter have little relevance nowadays, the dangers in overspending for software packages is applicable to current IT spending. Keller documents how the three categories of business software — internal custom, external custom, and prepackaged — have consumed increasing investments up until 2002. He concludes from this that, "Even though companies were buying more software packages, they still spent more money on the customized labor required to make them work." (page 32). However, internal custom software (developed in-house by staff and/or contractors) is not equivalent to customization of prepackaged software purchased externally and brought in-house. Thus, his conclusion is only defensible if it is based upon (uncited) evidence documenting that customization of prepackaged systems also attracted greater investment. In fact, without knowing what percentage of the "internal custom" work refers to customization of packages, it's impossible to draw any sure conclusions from this data.
Nonetheless, the author's primary point is clear and substantiated. The deceleration in IT spending was inevitable, partly because the ridiculous rate of growth could not be sustained, and also because of mounting horror stories indicating how much money big enterprise was shoveling down the rat hole of ill-defined, pie-in-the-sky IT projects. The seemingly unending corporate mistakes looked even worse when the press began reporting how many hundreds of millions of dollars were being lost by firms that had put their faith in expensive and all-encompassing "solutions" to their information management needs.
While these juicy stories are interesting to read, the most valuable part of the book up to this point is the "Survival Guide for Buyers", at the end of the chapter. It offers useful advice to buyers and sellers. In the former case, it reminds buyers of the high risks involved in complex IT projects, the value of taking a long-term view and of regrouping if needed, and other important considerations. Sellers are advised to simplify complexity, to not overpromise, to choose their customers carefully, and focus on their success.
Keller goes on to discuss how the best measure of IT's value is productivity, and yet that same productivity is so difficult if not impossible to measure. One statement, however, struck me as rather odd: "Buyers have begun to understand that the larger consideration for technology purchases is not if IT assists productivity, but rather when IT delivers benefits." (page 50). But if productivity is so difficult to measure, how can these businesses even determine whether it is assisting productivity or delivering any other benefits? He makes an interesting point that the bulk of productivity gains from IT were a result of the actual production of technology, and not the use of that which was produced. While it may be obvious to any experienced techie down in the trenches, some managers apparently need to be reminded that simply buying technology, without understanding what purpose it will serve and how it will do so, will likely be a waste of money. The price tag of a new tool does not indicate its eventual value to an organization; rather, its value is determined by how that tool is used over time, and for how long.
After detailing additional fiascoes resulting from the (attempted) implementations of large-scale software packages, the author continues with a brief look at IT spending averages. What is most notable is the dramatic differences among a few industries, such as telco services and financial services, which each spent more than triple and double the average of the industries, respectively. Yet these differences should not be that astonishing, given how telephony and financial data management are absolutely wedded to technology. The chapter wraps up with another Survival Guide, as does four more of the subsequent chapters.
In Chapter 5, Keller explores in greater detail much of the IT overspending, but this time focusing less on prepackaged software solutions, and more on other technology purchases, including software tools, database systems, servers, PCs, and other IT infrastructure. This information best serves as a reminder to managers to only purchase hardware and software that is needed, and not to make the common mistake of loading up on soon-to-be-outdated assets. Although not noted by the author, unused software and hardware — often referred to as "shelfware" and "overcapacity" — can quickly lose its resale value. He does, however, point up many of the most common mistakes in purchasing software licenses, and then goes on to provide greater details of systems management applications and analysis that are designed to help companies identify underutilized network capacity, unneeded routers, redundant database licenses, and other wasteful expenditures that can easily add up to significant amounts of lost money.
Yet Keller does correctly note that the first iteration of client/server computing, most popular during the late 1980s and early 1990s, proved to be more complex and costly than the centralized paradigm it was meant to replace. Web services are now enabling enterprises to distribute data and logic among many servers, all of which have much smaller acquisition and maintenance costs than the traditional mainframes necessitated by centralized computing in decades past. This switch to Internet architectures does not mean that companies should abandon their COBOL/mainframe systems, which still process over two thirds of the world's business transactions. One welcome theme throughout the book is Keller's understanding that shiny new technologies and applications do not necessarily result in progress on the part of their adopters. His is the first discussion of the pitfalls of Web services that I have yet seen.
Near the end of this sixth chapter, Keller discusses the effects of Microsoft, Linux and open-source software in driving down the costs of operating systems, database systems, Web servers, and financial applications. However, in a table comparing proprietary and open source software, he mixes company and product names in a manner that would be confusing to non-techies. This may have been the result of the publisher limiting the space available for the table. Given the importance and potential value of the information, the table should have been expanded to allow plenty of room for product and company names, as well as website addresses and additional categories — even if that meant consuming two full facing pages.
The discussion of offshoring, in Chapter 7, adds little new information or insight beyond what the typical business manager has seen in the plethora of articles on offshoring. The per-hour programmer costs of various labor pools — U.S. consultants, independent contractors, on-site and off-site Indian programmers, etc. — is compared between 2002 and 2006 (projected, no doubt). The figures suggest what economists have predicted ever since globalization of knowledge work first began, namely, a gradual leveling, proportional to each country's standard of living. Reductions of American salaries by a third will be accompanied by Chinese and Russian salaries increasing by half.
As a business consultant, Keller focuses on the cost reductions to the companies outsourcing their work to lower cost nations. As to the quality of the software developed overseas, he mentions anecdotal comments from three managers at separate companies, plus an explanation of how India has more development firms than America with high Software Engineering Institute (SEI) Capability Maturity Model (CMM) ratings. Anyone who has worked in a high CMM-rated environment, understands that it is no guarantee of code quality. Furthermore, he does not address the growing evidence of poor quality software been created by Indian firms, despite their high CMM ratings. On balance, he does mention some of the pitfalls, such as possible sabotage and unrest in these developing countries. He wisely urges companies to develop backup plans, in addition to actively managing projects sent overseas.
In the eighth chapter, the author identifies 14 spending drivers, most of which he maintains point towards flat or even declining IT budgets in the years ahead. The spending cutbacks will take place in four major areas: hardware, software, labor, and services. Keller foresees the greatest potential savings in those last two categories. These budgetary cutbacks are already being implemented at several of the companies cited by the author, with apparently little reduction in overall output. This spending slowdown is also reflected in reduced sales of hardware, software, and services, as well as renegotiations of license and maintenance contracts. The subsequent three chapters go over previous ground, but with more variations on the theme of reduced IT spending, interspersed with several examples from various corporations. The reader may get the sense that not much new information or recommendations are being offered, but instead that these four chapters are serving as filler, to beef up the size of the book. Otherwise, it would be more obvious that the book's usable contents could be boiled down into one meaty article.
In the book's epilogue, Keller gives his reasons as to why the lean years of today will likely continue for many years, even decades. While reading this final section, I had hoped that he would finally explain exactly how IT departments are going to deliver even more output, despite being provided even less financial input. Repeating the phrase "spend less to get more" is not sufficient to establish that such is possible, much less how it can be achieved. The closest that the author comes to actually spelling out how greater gains will be achieved, is his explanation of productions/deployment innovation: "The deployment of IT should combine business units providing the design and IT groups delivering the manufacture of the end products. When these groups work together, they can deliver a higher-quality product at a lower price." (page 215). This statement suggests that Keller lacks an accurate understanding of how software is designed. Business units lack the technical expertise to design software; they can and do, however, furnish the IT staff with the system requirements, which the programmers and architects then turn into system design and eventually code. If Keller is suggesting that the system users (i.e., the people in the business units) take over system design, then he does not realize what is involved in the design of software. In a nutshell, business and IT staff already work together, so there's no higher-quality-at-lower-price to be found in that suggestion. Moreover, asking businesspeople to design IT systems would be like asking homeowners to architect their houses.
Keller's primary thesis, that American IT could in the future produce more returns for less investment, has two primary components. The near term and likely long-term trend for declining corporate spending on IT, is well established in his book. In fact, one could argue that reduced IT spending is not something that American companies will adopt by choice, but instead will be forced upon them due to deflationary pressures, increased costs for natural resources, and declining ability to pass along cost increases to U.S. consumers falling further behind financially. But the flip side of his thesis, that companies will get even more results despite spending less money, is not nearly as well substantiated. Not a single one of the chapters in the book is devoted to demonstrating that this is happening, or will happen. Companies may be able to maintain current levels of service despite reduced funding; but greater results per dollar invested (i.e., efficiency) does not imply greater results on an absolute basis. As such, Keller's big claim noted earlier, is only half fulfilled.
The critical questions mentioned near the beginning of this review — concerning the proper role and funding of IT — are presented in the book couched in the language used by high-level business managers, who speak in vague terms about "technology" and "infrastructure", and yet have little or no real understanding of how it truly works, having spent their earlier years pursuing MBAs rather than programming computers. It could be argued that such general terminology must necessarily be used when discussing information technology among business managers. That may be true, but it does not lessen the dangers of fuzzy thinking and overly broad conclusions found in Keller's book and in the typical articles discussing IT purpose, strategy, and utilization. In particular, such excessively broad strokes, in my experience, not only mask the ignorance of the IT manager demanding miracles from their staff, but invariably increases the odds that upper management will be seduced by the handwaving consulting firms — and thus fall prey to the mistakes delineated by Keller.
Weighing in at 243 pages, Technology Paradise Lost is a quicker read than many other business books. Part of that is due to the unfortunate repetition of a few core ideas. Fortunately, the book has just enough tables, charts, and breakouts, to add some visual variety to the text. One inconsequential difference that I noticed, versus most technical books, is that none of the bullet points are right-justified, though the non-bulleted text is. This produces an odd visual effect for pages that mix bulleted and non-bulleted material. Also adding to the visual incongruity, is the publisher's decision to format chapter subheads in sentence case, and not title case.
Of all the inapt analogies in the book, its title is perhaps the most egregious. Alluding to John Milton's famous narrative poem, "technology paradise lost" implies that there was a time when IT resource usage was idyllic, if not perfect. Yet by Keller's own account, the misspending and failed projects, followed by financial discipline imposed by the outside world, are anything but heaven-sent. One cannot lose what has never been found.
As anyone in the IT industry knows, technological change can be rapid. This book would have been more valuable had it been published a couple of years earlier than 2004. The mounting IT failures were clearly evident by the late 1990s, and a book like this would have been far more timely had it come out before the year 2000. Understandably, however, it takes time (a couple of years, in fact) to get the typical nonfiction manuscript proposed, accepted, written, edited, and published.
In terms of errata, I found only two: "in which a computing unit becomes processor board" (page 99), and "Land's End" (page 141). These may have already been reported, and there may be others, but there's no way to tell, because at this time the publisher's website does not appear to have a way to report errata or review reported ones.
The book benefits from the author's clear writing style, no doubt honed from over two decades of creating articles, documents, and presentations intended for business managers. Keller does a solid job of utilizing real world statistics and examples, to backup his assessments.
Despite the repetition, sloppy analogies, and business-speak generality, Technology Paradise Lost offers a valid discussion of changes currently being experienced by the American IT industry as it grudgingly recovers from the Internet boom and bust. The book may be of value to IT managers who, for whatever reason, are ignorant of the obvious transformations that are taking place. Yet, any IT industry participant who devotes even a modicum of time to monitoring the latest developments and trends, should be well aware of IT budget trimming, offshoring, open source software, and other cost-saving methods. Otherwise, to be so out of touch with reality would be inexcusable. On the other hand, that was one of the primary symptoms before and during the widespread dot-com insanity, and could easily account for any beliefs in its imminent return.