Although I consider writing to be good or bad, good and bad suggest subjective or even moral judgment. To be more objective, Joseph M. Williams and other authorities state that we should worry more about whether writing is effective than whether it is good.
Effectiveness can be defined as a ratio of what the reader gains vs how hard the reader must work. If writing is effective, the reader gains a lot per unit work. In technical writing, much of what is gained consists of information or understanding about some system (an organism, a process, a problem, a technique, etc). In technical writing, a direct and lean style is almost always easier to understand than other styles. Some authorities suggest that the most effective technical writing is transparent, i.e., the reader obtains information without noticing the writing style at all.
In other kinds of prose, what is gained can be more than information about a system. Prose can also offer valuable information about how the system relates to some broader view or some narrower, personal view. These broader or narrower views may benefit from a writing style that is more than direct and lean; complex and nuanced ideas may benefit from a more complex and nuanced writing style. Here we also recognize that some of what the reader gains is from the style itself. In other words, the style is more than a transparent window through which the reader looks to obtain information. Style is part of the information.
Scientists too often write in an ineffective academic style. The remedy is to write more directly, concisely, and vigorously. But this doesn't have to produce transparent, invisible prose. Certainly, transparent and invisible prose is superior to the inflated kind, but part of the scientific information, especially what the author thinks about the information, cannot be conveyed with only transparent, objective words. The default style may be direct and lean, but the author should be willing to deviate from that, to add complexity and surface to the style, so as to more effectively convey information about who he or she is and what he or she thinks about the research.
Finally, even the most direct and concise scientific prose is unlikely to be transparent. Readers, at least those who pay attention*, will notice and appreciate this prose because it differs so much from the more common, inflated kind.
*As Kingsley Amis stated, readers who pay attention are the only kind worth bothering about.