You have 1 article left to read this month before you need to register a free LeadDev.com account.
Estimated reading time: 4 minutes
More than 71% of experienced software developers say their language migration decisions were influenced more by industry hype than by proven outcomes.
That’s according to a recent HostingAdvice.com survey of 500 developers who’ve migrated production systems in the past five years.
In an industry that prides itself on being data driven, the findings show that software engineers can be just as susceptible to industry hype and FOMO when it comes to the latest programming languages and frameworks.
As a result, only half of the developers surveyed said their migrations were worth it, and 58% said it introduced new technical debt.
Hype cycle
This sort of hype is prevalent in the world of front-end development, where Svelte gathered buzz as a lightweight alternative to React years ago, but has spawned countless alternatives, such as NueJS. Similar cycles have played out with AngularJS giving way to React, and Bootstrap losing mindshare to Tailwind, and so on.
“I’m not surprised by this at all,” says Joe Warnimont, senior analyst at HostingAdvice.com. “Isn’t everything driven by hype?”
Language migrations are happening frequently too, especially within large companies. In the past five years, 71% of teams have done over two migrations, and 38% have done over three.
Avoiding this age-old trap is tricky. Making the right decisions requires looking out for the warning signs of hype-driven migrations, learning from others through case studies, and assessing usage metrics.
Your inbox, upgraded.
Receive weekly engineering insights to level up your leadership approach.
An AI-fueled spike
By making it easier than before, AI coding tools are fueling this constant technical churn, with 82% saying that AI tools made the migration process either significantly or somewhat easier.
As a result, AI-assisted tools, like GitHub Copilot or automated transpilers that convert code written in one language to another – such as TransCoder AI, Emscripten, Babel, or CodeWhisperer – were used by 87% for migrations.
“Real-world developers and IT professionals are seeing positive outcomes when using AI during migrations,” says Warnimont. For example, consultancy Future Processing saw a 40% time savings using AI to migrate a client app from Angular to React.
“AI is truly good at interpreting legacy code and creating natural language specifications,” says Justin Reock, deputy CTO at developer productivity company DX. He points to Morgan Stanley’s DevGen.AI, an internal refactoring tool that saved the company more than 280,000 hours of work.
More like this
No regrets
“I think most programming language migrations are driven by hopeful thinking rather than proven outcomes,” says Charity Majors, CTO at Honeycomb. The broader side effects, like code maintenance or upskilling requirements, typically trump pure technical considerations, she adds.
She recalls migrating an API from Ruby to Go while at Parse in 2015. “It was 100% worth doing, even though it took almost two years and was a massive pain in the ass. The backend team shipped zero product during the entire time,” Majors said. Although it was a significant overhaul, it ended up streamlining deployments and allowed the API to better handle concurrent requests.
Fast forward ten years, and AI can slim down such efforts. But could newfound AI abilities lead to unnecessary revisions? The rush to adopt hyped technologies, paired with the instantaneous means to do so, is a recipe for misjudged migrations.
“When an engineer gets excited about a language, they turn into a hammer looking for a nail,” says Majors. “Sometimes that nail exists, and the results can be glorious. Sometimes they’re actually just screws.”
To avoid regret, Warnimont recommends piloting small-scale migrations and reviewing public case studies. “Look at actual studies done on the results from each migration, including studies that include metrics like performance gains, integration issues, and training required.”
Take why Shopify migrated its CLI from Ruby to TypeScript and Node. The decision was made to keep pace with modern development practices in their community, eliminate complex native extensions, and support more consistent coding patterns. The story clearly demonstrates why migration was necessary and what the tradeoffs and outcomes were.
For others, it comes down to having the right metrics, managing expectations from the start, and building AI literacy to avoid introducing defects. “Enable engineers on prompting best practices and high-value use cases,” says DX’s Reock.
He also notes that AI isn’t the only option for migrations. According to Reock, OpenRewrite, an open-source tool for large-scale source code refactoring, is far more accurate and deterministic once a proper recipe has been written, and solves the same challenge.
There’s also the leadership element. According to Majors, migrations need an internal champion to drive them forward. “I think a lot of the bad rewrite decisions I’ve seen have been linked to this dynamic.”
Hype continues to usher in new technical migrations. But making it worth it will take upfront steps to assess end outcomes before migrating. Otherwise, you may find yourself in that 50% camp that regrets following the hype after all.
