How to read a technology trend without getting used by it

Every few years the industry agrees on a new centre of gravity, and the conversation acquires a peculiar quality: the technology is discussed as though adoption were a moral position rather than an engineering decision. Companies that have not adopted it are described as behind. Companies that have are described as leading, whether or not anything has improved.

The useful skill is not scepticism, which is just as lazy as enthusiasm and considerably less pleasant. It is the ability to separate three questions that get collapsed into one.

Three questions, not one

The first: is the technology real? Usually yes. Genuine capability shifts are rarely fabricated; they are just early. The second: is it operable at your scale, with your constraints, by your team? This is where most enthusiasm dies, and it is an entirely different question from the first. The third: does it address a problem you actually have, ranked against the other problems you actually have?

A technology can pass the first test brilliantly and fail the third completely. That combination is what produces the expensive adoption everyone regrets — the platform that works exactly as advertised, solving a problem that was ninth on the list.

Useful diagnostics

• Who is reporting the success stories, and what do they sell? Vendor case studies are evidence of what is possible, not of what is typical.
• What is the failure mode when it goes wrong, and can your organisation absorb it? Novel technology tends to have novel failure modes and no accumulated operational folklore.
• What does the hiring market look like? If the skill is scarce now, your maintenance cost in three years is a hiring problem, not a technical one.
• What is the exit? Every adoption decision should include an honest estimate of what leaving would cost.
• Is there a smaller version of this bet? Most trends can be tested against one real workload before they are adopted as a direction.

Being deliberately late is a strategy

There is a persistent assumption that early adoption is the aggressive position and waiting is the timid one. In practice the reverse is often true. Waiting eighteen months lets somebody else pay for the documentation, the operational patterns, the security findings and the hiring market — and the gap in capability you give up is usually smaller than the gap in cost you avoid.

The organisations that handle trends best are rarely the earliest. They are the ones that decided in advance what would have to be true before they moved, and then actually checked.