Thoughtworks Technology Radar Oct 2024 – From Coding Assistance to AI Evolution
infoq.com
infoq.com
Tools, languages and technology take their place after a significant share of people have adopted them, not before. Trying to steer industry towards certain tools and away from others, ignores the irrational choices that belie them, and how resistant we may sometimes be to purported technical fixes.
We are creatures of habit, and will break those habits on occasion, but ideally not because of an imbibed fear of losing out---the tools need to be proven in practice, not on promise.
Your perspective is biased to a particular point. The technology adoption bell curve (pioneers, early adopters, early majority, late majority, laggards) is a more instructive prism. Our propensity to assume risk is commensurate with both the size of the problem we're trying to solve and the lack of lower-risk options.
Pioneers aren't paying attention to tech radars, because brand-new projects find their earliest users directly, not via tech radars. The audience for tech radars are early adopters and early majority. Both are looking for validation from the rest of the market (i.e. seeking to avoid projects rated as "hold"/avoid), the difference being that early adopters are more risk-friendly (i.e. willing to try "assess") while the early majority is less risk friendly (i.e. looking for the "trial" recommendation).
At no point is a tech radar trying to get you to ignore your own evaluation cycle or to push projects ("adopt") that are not right for you. It's just a relatively organic marketing funnel to help projects with minimum proving to expand their audience, and for the wider market to connect to potential solutions, in exchange for the market developing appreciation for ThoughtWorks and thus serving as a potential marketing funnel for ThoughtWorks's primary consulting business.
My gripe is that it cements a certain way of thinking about how to best organise our stacks, to the degree that other companies are following suit by copying the 'radar' template without questioning its organising logic (why are languages/frameworks one category, for instance).
That's not to say we cannot categorise tools and techniques or inventorise new ones, but that we should not take the opinion of one entity to reflect industry as a whole.
As a nice counter example in this context, Zalando has tuned the quadrant categories to their specific way of working. Their categories are: Datastores, Data Management, Infrastructure, and Languages.
It almost feels like you're using irony to imply a connection between selling software consulting, and pumping-and-dumping tech trends. But I'm sure that's not your intention.
To me, an immediate improvement would be to display user issues and workarounds along the newly reported technologies. Or why someone tried it and then left if on the shelf. Possibly in context of a multi-team project.
Not a full write-up is needed here, but some practical user stories and anecdata alongside the current summary/ranking on the adoption radar would go a long way.
ThoughtWorks even provides open-source tooling to help you set this up.
For our second iteration we rewrote the entire radar frontend to better suit our needs. Using ChatGPT, Cursor, and Copilot, this was a breeze.
It is very much the opposite of keep-it-simple and use-what-is-proven.
I was going to say you're not alone, but for the purposes of identifying "what-is-proven" you can use the report to discount anything not labeled "adopt" as a starter.
A nicely-presented summary of various tech applications but I’m not sure we need a lighthouse anymore.
GraphQL for data products ...