Skip to content
February 10, 2015 / jpschimel

Data vs. Knowledge: a sorry letter from 1963

Thanks to Andy Hoffman for a column that brought to my attention this letter (Chaos in the Brickyard) that was published in Science in 1963. Already, the author, Dr. Bernard Forscher was concerned that science was increasingly focused on producing “bricks”–little pieces of information, rather than “Edifices”–real bodies of useful knowledge. That was in 1963 people!

Forscher Bricks vs. Edifices

Dr. Forscher wrote this as a parable, but the idea is directly analogous to my distinction between data, information, knowledge, and understanding. Ultimately science is about edifices: knowledge and understanding. Yet many scientists never do more than make bricks–individual nuggets of information.

I think Dr. Forscher was visionary: our reward systems increasingly skew toward rewarding bricks. The more we focus these systems on simple quantitative metrics, such as the Impact Factor of the journal a paper was published in, the more Dr. Forscher’s vision becomes true and the more we end up focused on the “bricks for brick-ness’ sake” rather than on the ultimate goal of building structures.

Reward systems should target direct measures of a scientists contribution to building real structures of knowledge, rather than just on their ability to produce bricks. That is something that peers can usually do easily, and sophisticated network mapping software can do with difficulty (e.g. Thompson Reuters Citation Laureates), but the simple metrics are mostly incapable of–they only measure bricks.

The concerns Dr. Forscher raised in 1963 have only become more extreme as has the importance of the reminder that to have real impact as a scientist, you need to force yourself and your trainees to aim for edifices. The ultimate goal of science is to advance human understanding. We build our great edifices of understanding out of individual pieces of information (bricks), but a brick’s value is only as great as its contribution to an edifice. A brick, by itself, on it’s own and isolated, remains a block of clay and straw. The papers that ultimately matter are those that offer more than just information–they offer real knowledge and insight.

Andy Hoffman’s column is available at: (http://chronicle.com/article/Isolated-Scholars-Making/151707/)

February 6, 2015 / jpschimel

These Things You Cannot Take

jpschimel:

This isn’t about “Writing Science,” rather it is the ultimate of personal writing by someone who I respect deeply for his science but equally for his humanity and eloquence in the face of adversity. I repost it because they are people who are close to me, a message that is equally so, and it shows a power of language. Alan, Diana, and Neva, the thoughts and wishes of all your friends are with you.

Originally posted on Alan Townsend's Blog:

My wife Diana was recently diagnosed with a brain tumor, the second brain tumor to hit our family in fifteen months. They are unconnected events – lightning striking twice unimaginably.  For the first one, still very much a part of our young daughter’s life, we shared a great deal here.  We may do the same for this latest challenge, or we may not.  It’s a bridge we have not yet crossed.  But on World Cancer Day, here are a few words.  

I’ve had a recurrent dream since last Friday.  I am afloat on an equatorial sea, a sweep of beach to my back.  Before me, lines of aquamarine rise from the depths in metronomic intervals as though the sea itself is but the rippling skin of some unseen leviathan.

The waves begin to build, each one just a touch higher than the last, and in the crystalline walls appear the faces of those…

View original 852 more words

January 26, 2015 / jpschimel

Language Pet Peeves #2: “Impact”

Here’s another word I want to add to my pet peeve list: “Impact.” It’s overused to the point of meaninglessness, and it often has little meaning anyhow.

I blame Thompson-Reuters for this over-use because they developed the “Impact Factor.” This, of course is a way of assessing the average citation levels for journals. As the impact factor has become the dominating metric of the publishing world, so has it emphasized the word “impact” as something to strive for. So, I see many people using “impact” in hopes of increasing their impact factor. Sorry, it doesn’t work that way.

I know some avowed language purists argue that “impact” should be restricted to a physical blow (its original meaning), but its figurative meaning “”the effective action of one thing or person upon another; the effect of such action; influence; impression” [OED] has been in use since at least the early 1800’s. Used well and carefully, it can be a great word.

No, my issues with “impact” are massive over-use and minimal clarity. When you try to express the impactfullness of your work by describing every effect and influence as an impact, the real impact of all your “impacts” is to deaden readers’ senses and reduce the impact of “impact.” Ok, that’s overkill, but you get the point.

More important though, is that “impact” is a fuzzy verb (Writing Science section 14.2, pg. 137). It sounds specific and concrete, but it isn’t—rather, it says there was an effect, but not what the effect was. That is a problem that goes way beyond mere overuse, which is a crime against linguistic aesthetics and style. Being unclear is failing in communication.

Consider sentences such as the following:

There was a strong impact of added ESX on the growth of Daphnia.”

The rate of GDH activity was impacted by altered salinity.”

Here, “impact” is a synonym for “effect,” and is devoid of content. Did ESX increase or decrease Daphnia growth? If ESX were presented as a potential toxin, I might infer that the effects were bad, since “impact” suggests a strong result, but the word itself doesn’t tell me—thus if fails in its mission. Instead, tell me what happened: “ESX substantially reduced growth” or “GDH activity was increased.”

You can sometimes use words like “influenced” or “altered” as an opening when the patterns of response were complex and you can’t capture the pattern with a single word, for example, “ESX altered the growth of Daphnia, initially increasing growth rate but later reducing it, such that overall yield was reduced.” But “impact” is a poor choice for that role, as it implies simple, direct, substantial action (e.g. an asteroid’s impact crater).

Remember the writer’s rule: “Show, don’t tell.” If you show us that there were interesting and significant influences and patterns in your results, you will convince us. Describing those influences as “impacts” to make them feel important won’t. Rather, it will feel like you’re trying too hard to speak for your results, which makes me suspicious. The role of language in science is to help the data speak for themselves. It’s to bring out and highlight what nature is showing us—not to try to impose a message upon nature.

“Impact” often feels like the author is trying to pump up their results, is heavily overused, and is surprisingly lacking in substance. That’s a trifecta that earns it this week’s place in Josh’s language pet peeves.

January 3, 2015 / jpschimel

Language Pet Peeves

This may become a recurring column, as even among experienced and knowledgeable scholars, there are some really common misuses of language. Here, as a start, are two that I find particularly frustrating, although I’m not particularly pedantic.

Principal vs. Principle.

Principal: The lead person.

Principle: A code of personal action (she has principles) or a fundamental truth (e.g. first principles).

I can’t say how many proposals I’ve seen where the investigators describe themselves as Principle investigators. Why can’t people remember that they are Principal investigators: the primary, important, person in charge. I certainly expect investigators to have principles, but a “Principle Investigator” is studying ethics, not science.

Combining words that should stay separate–nouns that should not be used as a verb:

Feedback: This word is only a noun. It may not be correctly used as a verb. Hence ecosystems can’t feedback on climate. The may feed back on climate. The combined form is only for the noun: a feedback between ecosystems and climate.

Uptake: This is also only a noun. You can take up something but you can’t uptake it. You can measure uptake, but this is now grammatically a noun.

Breakdown: Also not a verb. The verb is two words: break down, but a breakdown is an event. When your car broke down, it had a breakdown.

The pattern here is that these word pairs combine an action (feed, take, break) and a direction (up, down, back), the verb form leaves the words separate: the action stays distinct. Only when you make them into a compound noun do the words smash together.

January 2, 2015 / jpschimel

Condensing on a deadline, or Why justified margins are not justified

When trying to condense an 18-page draft into a 15-page limit, it’s not words you need to cut, but lines. If cutting a word doesn’t save you a line, it doesn’t achieve anything.

Particularly when time is pressing (which it almost always is when there’s a page limit), my approach is to first look for the low hanging fruit. Or in this case, the low hanging words; paragraphs with a few words dangling on the last line—words known as “orphans” in typography. If you can tighten the language and cut a few well chosen words, you can pull those orphans up into the body of a paragraph and save an entire line. It is hard to save a line when a paragraph’s last line is full; then you probably have to cut an entire sentence.

Cutting a single word, however, won’t necessarily open the space needed for an extra word, and won’t necessarily give the orphans a home in the body of the paragraph. Whether cutting one word will save a line depends on three things: 1) how big a word you cut, 2) how much extra space there is on the line already, and 3) how long the first word on the next line is. Cutting a short word on a full line that is followed by a long word isn’t going to save space.

For example, let’s look at the following paragraph (screen shots from Microsoft word, Times New Roman 12 pt. text):

Just p 1

I’m sure I can trim enough to make room for the orphan “individual monomers” in the body of the paragraph and so save a line.

One approach to condensing is to start at the beginning and find words to cut. Using this approach, the first word, I’d eliminate is “enzyme” in the second sentence: we’ve already identified that b-Galactosidase is an enzyme, so we don’t need to restate it.

Just p 2

But, this didn’t solve the problem! The first word on the following line, “exoglycosidase,” is too long to fit on the first line. The language is tighter, but I still have the orphans.

A quicker approach is to look for easy targets. The obvious one here is the 4th line, which ends with a big space. It should be easier to make enough space for an extra word. It also has an easy word to delete: “long” from “long polymers.” Polymers are by definition “long,” so “long” is an empty amplifier—an adjective that adds no real meaning (Writing Science pg. 164).

Just p 3

Bingo! That did it. By scanning the right hand margin, I was able to quickly identify a place where there was room I could use, and so speed the work of getting rid of the orphans and save myself a line. The alternative of working through word-by-word might have gotten me to the same place in the end, but it would have taken longer.

I was able to scan for the easy target because I formatted the paragraph as “Align Text Left” (the Microsoft term for what typographers call “Flush Left”) which leaves a ragged right edge. If I had switched this formatting to “Justified,” which forces the text to align along both the left and right margins, the original paragraph would have looked like:

Just p 4There are two problems here:

  1. I can’t see where there may be an easy place to cut. If you really like the tidy look of justified margins, you can edit in flush left and then switch to justified, but that raises the other problem with justified margins in word-processers.
  1. It’s hard to read. People don’t read letter-by-letter, or even word-by-word. We actually read by grabbing groups of small words at a time. The only time you have to stop for single words is for long, complex, or unusual words. To read quickly and smoothly, there shouldn’t be too much space between words and those spaces should be consistent. Word justifies margins by inserting extra spaces between words. But this can leave long and erratic gaps between words, making the text  hard to read.

We’re used to thinking that justified margins are pretty and professional-looking; after all books and magazines invariably use justified margins and they are wonderfully easy to read. True, but they use sophisticated typesetting systems that adjust spacing between words and even between letters within a word, and which hyphenate words when necessary to avoid creating excessive spaces. Microsoft Word can hyphenate, but how many people even know where to find the settings for it? If you hyphenate the original example paragraph, it looks like this:Just p 5

This doesn’t have the nasty gaps between words that the simple justified version does, but it didn’t eliminate orphans, and hyphenating technical words makes them harder to read.

If you’re producing a document using a sophisticated typography system, you can do proper justifying that adjusts space between letters as well as between words to get polished text and properly justified margins. But do any of us ever use those systems? It’s possible in LaTex, but even LyX, a popular LaTex-based word processor,  by default only adjusts inter-word, rather than inter-letter, spacing and so will be no better than MS Word at avoiding awkward spacing within paragraphs. In a Column in Fonts.com about justified type Ilene Strizver says “the wise designer will refrain from using justified type unless there’s a compelling reason to do so, and only when he or she has the time and flexibility to fine-tune the text.”

My Principal #1 is “It is the author’s job to make the reader’s job easy.” If you’re not using sophisticated typography, the best way to do that is to use flush left formatting. It will be easier to read and it will be much easier to work in if you need to condense to a page limit.

December 29, 2014 / jpschimel

Why journals sometimes are slow processing papers: look in the mirror

As a Chief Editor, I often get requests like: “We need a few more weeks to address the comments from reviewer #1? I have been travelling and haven’t been able to resolve with my co-authors the last issue the reviewer raised.” Such inquiries always raise the question of how I should respond.

When considering the individual request in isolation, the answer should obviously be “of course.” It isn’t much time and I’d much rather authors get the paper right. This is a service job, and my responsibility is to support the community. Supporting authors makes being an editor satisfying, makes papers better, and it strengthens the journal.

But prospective authors decide where to submit papers based, in part, on the time it takes a journal to process papers—and the statistics available to assess that are the times from initial submission to acceptance and publication. If authors perceive a journal’s metrics to be poor and processing to be slow, they may send their most exciting work elsewhere. So approving an individual request for extra time weakens the journal.

I have been anointed with the responsibility of being one of the Chief Editors for the leading journal in my field and have the obligation to pass it on at least as well respected as I received it. So maintaining the statistics is important, but at what cost to serving individual authors’ needs? Should I be a hard-ass to ensure our statistics stay good? Should I act like an airline—close airplanes’ doors on time even when there are passengers sprinting through the airport connecting from a flight that just landed, even when the winds are favorable and the flight will still land over ½ hour early? If I had to chose the statistics over the sprinting passengers, I’d rather quit.

I will always prioritize actual individuals over impersonal statistics, I accepted the work of being an editor to serve my colleagues, even those I’ve never met. I have and always will give extra time to authors and reviewers who ask reasonably. I once even deliberately held a manuscript in limbo for the better part of a year to serve the needs of a student—it was clear that with a major revision it could be a nice paper, but this person was moving to a new position and wouldn’t be able to get that done for a long time, but for institutional politics worried that if the paper were rejected, it could jeopardize their position. I agreed to hold the paper in limbo until the new version was ready. When the time came, I declined the original version and slotted the “new” version back into the system—it was published with little fuss following a quick and easy review. Time-to-acceptance for published papers stayed good, but that one case put a dent in our time-to-decline stats. Would you have me treat you otherwise?

Potential authors, however, don’t see these interactions—they just see the overall stats and assume that if the numbers are bad it’s because the “journal” isn’t being efficient.

I admit that there are times when I have been laggard about some editorial work; for reasons including work overload, proposal deadlines, even vacation and just life–journals are not inanimate entities but teams of people who usually have other responsibilities.

Typically, however, what takes the longest time in the processing cycle is rarely me as editor. It’s getting people out in the reviewer community to respond to requests to review, and then to deliver, on time, reviews they agrees to do.

I am always amazed at how many scientists simply ignore requests to review. That can eat a week or more as editors wait, send out a reminder about the request, and then wait a little more to make sure the potential reviewer got the request. Is it really that hard to just hit “decline” if you’re not going to do the review? Do you have any clue how much it would smooth and speed things for everyone if we all could just remember to do that?

And then many of us can be terrible about getting reviews done once we’ve agreed to do them—I regularly have to send multiple personal reminders about due reviews. I often find myself on the horns of a dilemma, balancing the chances that one more round of nagging and guilt will kick loose a useful review vs. the chances that I’d be better off abandoning hope and starting the process over again trying to get someone to agree to do a review and then to deliver it on time. If I go that route I could, however, easily be back at the nagging and pleading stage, but just a month or more later.

So in discussing journal speed, please consider several things:

  1. Are you pushing editors to be ever more hard-assed about deadlines and being unsympathetic to your individual needs?
  2. Are you thinking about the journal as some faceless thing, rather than as us, your colleagues down the hall, and even you, yourself? Journals are based on peer review, after all, and if peers are slow, journals will be too.
  3. If you want journals to be fast in processing your papers, you must equally be fast in doing reviews for others.

When you gripe or hear someone else griping about how long it takes a journal to process your manuscript, remember the old saying from the comic strip Pogo: “We have met the enemy and he is us.”

January 28, 2014 / jpschimel

Ten Thousand Posters Invade the Moscone Center: Most Miss Their Target.

Invasions of the U.S. happen on Sundays in December. But this one you can see coming two airports away: people bearing tubes, some simple cardboard, some neon plastic; all heading for the Moscone Center to present posters at the American Geophysical Union annual conference.

Most of this wave of invaders will fail to achieve critical goals because they misunderstand their medium and the nature of their target, just as did the first wave of Sunday invaders in 1941. The Japanese used state-of-the-art airplanes to bomb obsolescent battleships instead of the fuel bunkers and support facilities that allowed a modern fleet to fight. AGU poster presenters largely seem to believe that they are presenting written communication to a patient audience.

But a poster is something you look at, not something you read: remember the definition of a poster is “a large printed picture suitable for decorative display.” A poster is visual, rather than written communication.

And a poster presenter’s target audience is the most impatient a scientific presenter will ever face. We’re way worse than Nature editors or proposal reviewers.

We’re impatient because there isn’t much time in a poster session and there is vast competition for our attention. At AGU, there are there almost 2 miles of posters each day and there’s free beer and friends we haven’t seen since the last meeting. We’re saturated with information and late nights.

I can’t assimilate all the information I see at any conference. I have neither the ability nor the patience to try to soak up all there is on the walls of posters. I must pick and choose which ones I invest my time on. Is this poster going to offer a useful new insight or a new friend or colleague? It takes me between 2 and 15 seconds to evaluate the balance of attraction (interesting topic, visually engaging, etc.) vs. repulsion (complexity, clutter) and decide whether to engage with the poster and the presenter.

I’ve seen a few “perfect posters” in my life. One was so clean that 25 years later, I can still reproduce the model of microbial growth and colonization it developed (thank you, Doug Caldwell). Another I don’t remember as being “perfect” visually, but it presented a data set so intriguing it started a conversation that led to a new collaboration and a manuscript that is now in press (thank you, Xiaofeng Xu). The others I remember were for a single figure, graph or conceptual model. Always, it is the visual elements or the presenter that engages; never the text. And always because the poster focused on a simple clear point that I could absorb quickly and use to launch a conversation with the presenter.

A poster is more like the slides you use to support a talk, than a paper that a reader should be able to work through independently. The visuals are the heart of it, and the text is just the glue to hold them together and to give enough context that someone can get the essential story without the presenter to walk them through it. It isn’t a paper, where you expect to read and contemplate the complex thoughts and arguments.

If I see a poster that is a wall of words, with paragraphs of text, particularly in normal type-sized type, and cluttered with too many data panels I will do one of two things: A) most likely, walk on by, figuring that trying to work through the material just won’t be worth my effort. Or, B) if the topic looks particularly relevant to me, the presenter is someone I know or comes from a group I respect, or perhaps even looks like a student who needs someone to talk to them, I’ll stop and ask “What’s the story?” I’d much rather hear it from the presenter than try to read it with her hovering next to me anyhow. Standing on hard floors makes my knees hurt and a good conversation distracts from the pain. The one thing I will almost never do is read the damned thing. So why put all those words there in the first place?

The key to producing a good poster is the same as for all communication—understand the medium and the audience. In the case of a poster, the medium is visual and the audience is impatient. That leads to several straightforward, but often ignored, rules, rules that are different than for written communication or a patient audience.

1. Your primary goal: Draw people into conversation so you can tell them about your work.
2. Title: Short, catchy—something that will draw a passer-by to talk to you.
3. Story: Simple. It must highlight the critical points: What is the problem? What is your question? What are the key results? What did you learn? Avoid details—we can get them from you or ignore them, and aren’t likely to remember them from reading the poster.
4. Text: Ideally, no text paragraphs. Truly, not one. Bullet lists, single sentences, short notes. Things that you absorb in a glance.
5. Flow: Ideas should emerge as your eyes flow across the poster from left to right and top to bottom.
6. Type size: Big. Nothing under 18 pt. and 24 is better.
7. Tables: Bad—they present information, but not insight. You have to read a table and think about it a lot to figure out what the point is.
8. Figures: Good—they show patterns and trends and so are more powerful for getting across the important insights.
9. Simple cartoons and schematics: Great. They are eye-catching and can communicate complex concepts powerfully.

These are rules, not principles, so you will likely violate some of them, but if you violate too many, people will ignore your poster. There is much more that could be said to illustrate these points, but this is a blog post, not a new book on visual design. David Schultz has a nice, if short, chapter on posters in his excellent “Eloquent Science.” Angelika Hofmann has a longer, chapter in her book “Scientific Writing and Communication,” which is also good.

As long as you construct a poster that reflects that it is visual communication for an impatient audience you will be likely to hit your target and achieve your goal of communicating elements of your work to the target audience.

Follow

Get every new post delivered to your Inbox.

Join 73 other followers