I think so. For productive office jobs, a desktop is still a must have. Audits, quality assurance, production planning, sales planning, reclamations, presentations, any form of computer aided engineering, accounting... just random examples on top of my head.
Sure, there could theoretically be a a webapp that can do all this, but there isn't. A penalty is that your QA SaaS service needs to interface employee information and maybe access article information living inside your ERP. Problems start to occur here.
Apps on the web or phone are good for any workflow where your input is basically "yes/no". I think html can be a good compromise for UIs, but that isn't enough to replace desktop software.
I don't see many webapps in any production critical process, they only pop up for some administrative tasks and even that is limited. Maybe you are an IT consultant with everything in the cloud. Or maybe you deliver food or offer transport and orders are managed in an online service. That works, but widespread applications are limited.
I don't think users care how software is deployed. (ie: does it live on the web or in their desktop?) But what looks dead to me is desktop-specific development.
With modern UI frameworks and Electron, you can deliver an app that looks "desktop" to the user but is written with standard, simple web technologies. Those apps interact with your peripherals and are indistinguishable from desktop apps.
The advantage to letting go of proprietary "desktop" software is that a single codebase can actually look and work the same on PCs, Macs, and Linux -- something no desktop provider has given us.
I mean for some applications you're going to gave to have a desktop app, particularly anything that communicates with external devices or most OS related components.
Although the ways people will convince themselves otherwise are pretty comical. I had written a desktop app that communicated with a card scanning device via a simple protocol over a serial port. Someone proceeded to tell me that desktop apps in 201X are bad. They told me the preferred alternative was to write a webserver that did the serial port communication and also served a web based UI which seemed ridiculous at best.
Sure, there could theoretically be a a webapp that can do all this, but there isn't. A penalty is that your QA SaaS service needs to interface employee information and maybe access article information living inside your ERP. Problems start to occur here.
Apps on the web or phone are good for any workflow where your input is basically "yes/no". I think html can be a good compromise for UIs, but that isn't enough to replace desktop software.
I don't see many webapps in any production critical process, they only pop up for some administrative tasks and even that is limited. Maybe you are an IT consultant with everything in the cloud. Or maybe you deliver food or offer transport and orders are managed in an online service. That works, but widespread applications are limited.