The core of the discussion seems to be centered on two things:
1. Formal science writing, as appears in peer-reviewed journals, has a particular style based on precise language and is consequently difficult or uninteresting to read for the majority of non-scientists.
2. Some science writing, typically that directed specifically at non-scientists, is very entertaining writing, and ranks as excellent writing by most (subjective) scales of measurement.
The problem, as much as there is a problem here, is that in order to make science exciting to a large number of non-scientists, certain conventions of scientific writing must be abandonded. I'd like to argue here that we (scientists) should not abandon these conventions, only relax them slightly, and that such highly-precise language can still be highly entertaining.
The real problem, in my opinion, is that people are easily distracted by nearly-meaningless words and phrases, which (as a result) have been carefully excluded from the normal language of science. Non-science writing is full of these words, phrases and conventions, and once someone (e.g. me) has read a large enough number of good scientific articles, the popular press becomes a morass of un-cited sources, 'spin', and propaganda. If I must adopt these morally-questionable writing styles in order to popularize my work, or just to have a conversation about my work with a non-scientist, then I think I will not bother. Honesty is probably the highest ideal of any scientist - which is why an honest mistake is never* reflected too poorly upon a scientist.
Returning to point 1, above, I have seen (even in my short time as a scientist) a shift in scientific writing style. It's easy to see a large shift in style over the past 100 years - some journals (and journal-archive sites like JSTOR) maintain electronic archives of scanned-in articles stretching back into the 1800's. The shift over the last, say, 25 years (yes, I often read papers from at least that long ago) seems to have been away from passive voice ("experiment X was performed") to active voice ("we performed experiment X"). It's a relatively minor shift, I think, but it does seem to make scientific writing a little easier to read. I also like it because passive voice often gives the impression that the happy little nocturnal lab-elves did the actual work, and the authors of the study just typed the results into their computers in the morning. Active voice makes it clear that the nocturnal night-elves succumbed to the lab pest-control fumigants decades ago, and anybody scribbling in a note-book at 2:00am is most likely a graduate student.
I don't think scientific writing conventions should be sacrificed to reach a broader audience. Some efforts can be made in making some work less dry-seeming, particularly where the target audience is extremely unlikely to conduct replication of the experiments, and hence doesn't need nearly the usual level of detail about materials and methods. Interestingly, Nature has just announced expanded M&M sections in future publications, driving them in the opposite direction - a shift probably motivated by the fact their audience is almost entirely actual working scientists, who are interested in those methodological details. In any case, if one is writing about one's own work for the local paper, it seems unwise to chew up column-space with careful descriptions of the addresses of the manufacturers of one's microcentrifuge tubes.
I have seen many examples of easy (dare I say "fun"?) to read science writing in proper, peer-reviewed articles. My favourite example is the entire collected writings of Susumu Ohno. I haven't read all of his stuff, but one example I'll show came from a 1974 paper (1). Page 23:
This then is the process of diploidization. Since quantitatively one of each of these newly created pairs of gene loci is redundant, it is free to accumulate formerly forbidden mutations. As a result, it may be silenced. Alternatively, on rare occasions, it may as mentioned in Chapter 1, emerge triumphant as a new gene locus with a previously nonexistent function. Thus a diploidized tetraploid becomes endowed with a far greater possibility than its diploid ancester to sponsor a large evolutional change.
Notice the use of language! A gene emerges triumphant! An organism becomes endowed with a great possibility! Mutations are no longer forbidden! I love this stuff, and it's not at all imprecise, or conveying the wrong impression.
Non-scientific writing is typically imprecise and laden with unnecessary descriptors. I've been marking undergraduate essays for the last week, and I have marked a great many undergraduate lab reports over the last two years, so much so that I now find myself resisting the urge to circle offending words - like "extremely", "very", or "critical" - with a red pen when I see them in books or newspapers. These extraneous modifiers now serve only to annoy me - I like to read writing that has been pared down to the minimum information, as I consider a well-turned minimalist phrase the height of "elegant"; high praise in scientific writing, from many sources.
Replacing 'jargon' with words of ambiguous or multiple meanings when converting a scientific paper to "popular" reduces transmitted information. Adding modifiers usually transmits no additional information, except perhaps little warning flags that say "this author is hyping!". I do not think that 'dumbing-down' a phrase like
"We found a positive relationship between body size and total number of eggs produced, as well as body size and number of eggs per egg mass."is equivalent to
"We found that the larger the snail, the more eggs it produced and the more eggs were in each egg mass.".(2)
The second sentence is superficially very similar to the first, but omits some information and unnecessarily channels the reader's mind into one direction of movement along a variable's axis. In this case, the distinction between two separate correlations (body size & egg number and body size & number of eggs per egg mass) is collapsed into one three-way correlation (body size & egg number & egg number per egg mass), which implies non-independence of several other variables like number of egg masses. The reader's mind gets channeled into moving only upwards on the body size variable by the "...larger the snail, the more..." statement - the first sentence more accurately conveys that this relationship is properly linear, and one may examine the relationships between X and Y at any point along the line.
Another idea that came from the discussion at Pharyngula concerns motivation - what incentive is there for a scientist to write for a non-scientific audience? I think it would be great, personally, if I could convey my work, and how I feel about it, to my non-scientist friends and relatives (especially relatives - my aunt would prefer I work on "something cute", even after I pointed out that work on "cute" animals typically involves invasive surgery). I think some modest recognition in the popular press, perhaps at the level of recognition one gets for writing a well-recieved letter to the editor of a major newspaper, would also be quite wonderful. Any improvement in 'society's' opinion of scientists and/or graduate students would benefit me, even if only in a small way. Certainly there are possible reactions more pleasant than "Oh, you're a grad student. What are you doing for the summer? Do you have to get a job?"**.
Some incentives were proposed at Pharyngula, including a high-profile prize for excellence in communication of science to non-scientists. I think this is a good idea, but that if implemented will have little effect on generating the next great scientist/writer, and zero effect on convincing normal scientists to practice writing for a non-scientific audience, for the same reasons that the Nobel prizes have almost no effect on generating excellence in research (nobody works on their thesis*** because they think they'll get a Nobel 20 years later) and no effect on normal researchers (I have no chance of a Nobel - so I'm not going to work harder just because they exist). I still think it's a good idea, for the same reason I think the Nobels are a good idea - some research, and some scientific writing, is so good it deserves very broad recognition.
If there are to be incentives for scientists to popularize their work, they need to apply to most working scientists, not only the extreme high-performers. The Public Library of Science journals have started a modest incentive scheme along these lines, though perhaps it was not intended as an incentive scheme when devised. Each article is presented with a short summary article to introduce the work to non-specialists, which often translates into an article that popularizes the work for the general public. The people who write these introductory articles are not the authors of the research paper - but they are working scientists, often people who might have reviewed the work in question - in other words, the "peers" of "peer-review". I don't know how much prestige is attached to these popularizations - probably very little at this point, though of course authors are free to include those articles on their C.V.s. I know of one case of a similar popularizing article, in Nature, written by an assistant professor I know quite well and have worked with in the past, that at least generated some interest among his colleagues at his home university.
Currently, these popularizing articles are written by other experts in the field. To popularize one's own work, which is a worthy goal, similar schemes need to be implemented in which scientists write their own editorial copy. Normal scientific conventions against 'spin' or 'hype' need to be enforced, and these could still be peer-reviewed, maintaining their stature in the eyes of scientists. There are a few scientists who do this well, but there are thousands of scientists who have never formally tried.
I'd like to propose an idea now, at the end of this overly-long blog post. I'd like to see a peer-reviewed journal, perhaps associated with a normal scientific journal, that publishes short descriptions (say, one page) of articles appearing in other journals. This popularizer journal would be available to a broader audience than the core scientific journal(s) because it would cost much less, and perhaps be given free of charge to public libraries and the like. I emphasize public libraries because university and other institutional libraries already have access to the core journals, and their users are often scientists who don't want to read popularizations. I'm imagining one journal that publishes popularizations of a number of associated core journals, thus reaching a broad audience and including a wide diversity of subjects. What do you think?
*well, at least such honest mistakes do not have serious consequences for working scientists in my happy little fantasy-land of perfect scientific harmony. How long until my candidacy exam, again?
**yes, I actually got that response once. I forget from whom.
***many Nobels are awarded for work conducted in graduate school.
(1) Ohno S. 1974. Protochordata, Cyclostomata and Pisces. Animal Cytogenetics 4(1): 1-92.
(2) Mona Albano, comment #20, Pharyngula post "Would Chuck Terhark like a job writing science abstracts?" February 17, 2007.