W51 - Terminal Convergence Changes Seen from D2

Last week I reviewed some text and videos from D2, digested them over the week, and will share some of my shallow perspectives on this edition of D2.

This D2's theme is "Fusion Future." For the first time in seventeen years, D2 upgraded its positioning from a front-end technology conference to an end-device technology conference, which carries certain historical particularities.

Because D2 had no keynote, it went straight into two days of vertical-domain talks. Judging only from the program, it was essentially a live version of our "Mobile Front-End Monthly," without much that was extraordinary. I tried to draw some conclusions from the change in theme and interpret the historical particularity as I understand it.

The "fusion" in the theme refers to the convergence of endpoints.Around 2017 the industry proposed and popularized the first convergence, commonly called the "big front-end," which has continued to the present. That convergence happened at the organizational level; it did not deeply integrate engineers' skill models or role ladders. Web and native teams still worked separately, and a single requirement might still need three different teams to support it. The 2022 convergence goes further, transforming the big front-end into an endpoint. The organizer made a bold move this year by merging Web and Native job codes. That implies Web and Native skill models and promotion models are now largely aligned, embedding an expectation for multi-skilled roles. I view this change as transformative because, in the medium to long term, it reshapes engineers and restructures organizations.

On the technological transformation.The hybrid path is long and difficult, but both internal and external historical experience show that, under a long-term view, hybrid is promising and, under certain historical conditions, inevitable. The direction is clear; what’s uncertain is the technical route—whether the end state will be an incremental enhancement approach or a generational upgrade. Many cross-platform technologies are popular today, but none yet stands as an absolute benchmark in depth and maturity. Until concrete standards emerge, I think we are still in an exploratory ascent. Broadly speaking, the technical evolution is trending toward implementing technology deeper in the stack: from the client layer, to container layers built on webview, to self-rendering frameworks like Flutter, and now to JS engines like QKing shared at D2. Internally at Jinfu, the progression from Titans to Neo to Recce also supports this evolutionary path.

On the transformation of people.First, lower development barriers and converging workflows reduce the need for collaboration. For collaboration optimization, one approach is to improve efficiency; another is to reduce the number of participants who need to collaborate. Endpoint convergence takes the latter route. With major advances in infrastructure and frameworks, a requirement that once needed multi-end collaboration can become a single-developer closed-loop delivery. Second, there are higher and broader fundamental skill requirements. Final convergence is not a bet on a single tech stack; we should emphasize engineers’ core competencies. In their skill toolbox, capabilities in programming, engineering, and architecture will be required more comprehensively.

On why initiate change at this moment.First, the survival environment has changed. The company is the same company, but the internet is no longer the same. Five years ago, if a company considered anyone who arrived before 10 o’clock late, would you choose to work there? Today you’d realize that’s trivial. Second, market expectations have shifted. True internet experiences, Apple’s XR, the metaverse—many visions seem fanciful or unreliable. You can dismiss some of them, but when the whole world yearns for a new platform, the rising force behind it is unstoppable. In this context, whoever figures out endpoint development cost and efficiency before the new platform scales will gain a decisive head start.