The Three Shades of Legacy
(Legacy Series 2/3)
My series on legacy systems will show how the adherence to old tech, processes and skills is the major threat for companies and economies. In the wake of breakthrough innovation such as blockchain or AI, legacy-strapped China is vying to dominate global business and finance. In three essays I will examine what policymakers and managers must do to capture the promise of the digital age.
What do many banking giants, the European space agency, and 19th century Irish whisky distillers have in common? They all fell victim to legacy systems.
Perils and remedies depend strongly on the type of legacy in place. Each heritage system falls into one of three major categories: Type 1 is creeping legacy and results from constant incremental evolution. Hardware becomes slow and software incompatible. The second type is industry-paralyzing legacy. Breakthroughs with the heft to turn an entire industry upside down occur suddenly and render existing systems inferior. Finally, Type 3 legacy occurs when carrier-technologies of the new techno-economic paradigm alter the back-end mechanisms throughout the economy and thus the very nature of competition.
Type 1 legacy – Bugs, outages, and other insidious trouble
In most cases creeping legacy stymies speed and incurs mind-boggling maintenance costs. Glitches and bugs are another result of dated set-ups. Whether it is Natwest customers being locked out of their banking system, 275,000 failed HSBC payments, or the delay of thousands of flights with Delta Airlines – the root cause lies with systems that grew obsolete incrementally.
Type 1 legacy can have even much more dire consequences. Ariane 5 Flight 501 of the European Space Agency exploded because of redundant code. Software written for Ariane 4 was used and eventually turned out incompatible with newer components.
Disasters caused by legacy systems even transcend the boundaries of large organizations and have cataclysmic effects on the entire economy. Banks in particular suffer from data fragmentation. Information is trapped in silos, and there is no overview, no master-file, no joint database. As The Economist reported, in the years running up to the financial crisis banks were incapable of tracking their risk exposure, a major enabler of the 2008 financial meltdown.
How to combat Type 1 legacy? Creeping legacy is the most easily manageable of the three types. Moore’s law helps to gauge the speed of future chip density. Software updates can be planned as they appear in stable cycles. And subscription and leasing models ensure constant access to the latest technology. Cost-benefit analyses are simple with creeping legacy. Even non-IT machinery gains resulting from more efficient tools are easily comparable with the impending investment.
There is nothing complex in this category, but companies have to set aside the necessary resources without compromise. Because if they don’t, technical debt will accumulate and will become more dangerous and more expensive to remedy every day.
Type 2 legacy – Industry-specific legacy
Nobody doubts the benefits of upgrading creeping legacy; discussions primarily revolve around the costs. Industry-paralyzing legacy, on the other side, evokes tribal support. Because the innovation is game-changing instead of merely efficiency-enhancing and because it only affects one industry, its benefits are dismissed and customer needs misinterpreted. 19th century Irish whisky makers are a case in point. They refused to switch to so-called column stills, which would have saved them 50-70% of production costs. They erroneously claimed customers preferred the pot still taste. The unstoppable ascent of the Scotch – and a crisis of the Irish – distilleries began.
The reluctance to switch systems results from the stellar leap in cost and complexity from type 1 to type 2 legacy. Have you ever wondered why credit-card-pioneer America was among the last to introduce chip-cards? Because they had invested more than anybody else in the build-up of the infrastructure. Only when credit card fraud in America exploded did the major card networks mandate banks to shift from mag-stripe to chip technology. But this meant that cards had to be re-issued, new suppliers found, terminals replaced, deals with retailers re-negotiated, and the public educated. Fighting type 2 legacy means a massive commitment and no manager will make this decision easily.
Today’s most important industry-specific shifts occur in computing, the major purveyor of the digital paradigm. Coding languages are a hybrid of type 1 and 2 legacy and thus among the most severe heritage problems. There is classic upgrading and there are paradigmatic switches such as object-oriented programming. Old coding languages exponentially drive up software entropy, tear open security holes, and impede modifications and new interfaces. COBOL and Fortran, both developed in the 1950-ies, are the most widely spread legacy languages. Even Java is more than 25 years old.
But sometimes there is innovation that allows you to stick to your systems and still enjoy the benefits of new tech. In computing, application programming interfaces (APIs) oust point-to-point coding and middleware, the traditional way of gluing new applications to old back-ends. APIs are powerful docking stations for other software. They get rid of large-scale coding projects and eliminate fragile dependencies from point-to-point links.
Microservices are yet another breakthrough. These mini-APIs offer smaller interfaces and are more frequently used for internal links. As great as these new techniques are, they make a manager’s decision more difficult. If you have a possibility to keep your old systems and evade an IT-transformation project, how can you argue to replace the old tech? And should you?
Breaking industry-paralyzing legacy also carries new opportunities. When the column still was introduced, it did not only slash production costs but made the whisky taste lighter and expanded the potential market. When chip cards replaced mag-stripes, contactless cards and mobile payment became possible, bringing payments to the phone. Europe’s new payment serviced directive (PSD2) hints at the potential of APIs. Mandating the API-opening of financial actors for each other, new services and revenue potentials sprung up for challengers as well as incumbents.
How to combat Type 2 legacy? The first question to ask with type 2 legacy is whether you need to completely exchange your systems – as would have been the case with the Irish whiskey makers or chip-based credit cards – or whether you can patch new technology onto old ones. In the former case, it is a very tough, but rather straightforward strategic decision based on the question: Do you believe that
If you have the option to stick new tech to old one, the discussion should be led more by your technical staff, not your CEO or business-oriented executives. As we will discuss in the next issue, using innovation such as APIs or microservices leads to integration debt and is fraught with its own complexity.
But by more long-term thinking, you can significantly reduce the necessity for fighting legacy to begin with. Regardless which industry you are talking about, you should keep the technological design as flexible and as modular as possible. In software, for example, a great way to prevent legacy problems is to utilize open source software, so you are not at the mercy of vendor.
Type 3 legacy – paradigmatic shifts
The technological theory tells us the Roman Empire fell because the Romans clung to an outdated and doomed system: slavery. Cheap and abundant workforce had a stultifying effect on the deployment of labor-saving innovation. It is why Romans fell short of scaling breakthrough innovation such as windmills and water wheels. They had invented them but continued to gorge on physical labor. When in the first century A.D. the empire’s conquest ebbed and the supply of slaves dried up, the economy collapsed. Such failures are the result of a chain of legacy systems. Not only was forced labor cheaper but old horse harnessing made transportation of heavy material slow and expensive, thus prohibiting the building of factories.
Today the digital paradigm is in full swing. The internet has laid the groundwork. And right now, three back-end breakthroughs are transforming the economy: cloud computing, blockchain, and AI.
Based on virtualization software, cloud computing enables access to IT resources from any place, any device, and at any time. It turns in-house server systems into last-era-legacy.
Blockchain, or distributed ledger technology, is the mechanism enabling cryptocurrencies such as bitcoin. Trusted third-party models become obsolete. Today its use cases far transcend the financial realm. In my book Blockchain Babel I label the technology “cheap trust.” Just as water was a key resource in the industrial revolution, blockchain-infused trust can fuel the digital revolution by replacing trusted third parties with algorithms.
Finally, AI tries to mimic the human brain, coming up with applications from self-driving cars to chatbots. With its sub-branches such as neural networks, machine learning, deep learning, and cognitive computing it has the power to transform almost any industry.
Recently the three technologies have become more than a figment of futurists’ imagination; combined they will unfurl their entire potential. The cloud decouples IT-resources from the end-user. This causes an increased need for secure connections and transactions, which is guaranteed by the blockchain. Blockchains keep not only secure, but permanent ledgers, just perfect for building the troves of data needed to feed AI algorithms.
Their amalgamation still lacks mass-roll-out, but pioneering initiatives break ground. The British government cooperates with Google’s Deep Mind and weds AI and blockchain to analyze health data.
How to combat Type 3 legacy?
The first and most crucial step is to exactly pinpoint which of your systems is affected by a new ground-breaking change. The advent of the blockchain, for example, does not make an entire core banking system obsolete. It does render certain old processes or software modules obsolete – say the reconciliation between different accounts – but you still need others such as your risk engine.
Once these crucial elements are identified, there is only one option: swapping them. After all they are mutually exclusive. Change always comes with a price tag. But not changing them comes with a bigger one.
In the next and last article of the legacy series I will look at technical debt as a financial instrument. Companies can consciously choose to live with old systems for a short time in order to get a heads up. It is a strategic tool. But they must not forget to repay it. And we will look at a wholistic way to handle technical debt.