General writing tips

Here are some useful tips to avoid common errors in your papers with word-choice or placement:

• Keep straight “affect” (verb) v.s. “effect” (noun).
• Remember that “data” is a plural (singular is “datum”).
• Use “fewer” when referring to something you can enumerate, but “less” when its a continuous substance (i.e., “fewer ice cubes”, but “less water”).
• Be careful with the placement of the word “only”. It should always appear immediately before the word or phrase its limiting.
• Watch out for homonyms. For example: “They’re” vs. “their” vs. “there” -or- “your” vs. “you’re”. Other common problems are “too” vs. “to” and “its” (possessive) vs. “it’s” (it is).
• When you use “this” or “these” you always want to follow it with a word to reattach it to the subject it refers to. Avoid a floating this.
• Avoid the word “very” whenever possible; a phrase is typically stronger without it.
• Use “which” when you want to give more details about the subject, but use “that” when you need to specify what topic you are referring to. A quick rule: if the sentence still makes sense without the phrase starting with which/that, then it should be “which.” If the sentence doesn’t make sense, it should be “that”. Quicker rule: if you put a comma before it, use “which”.
• Do not use single digit numbers unless they represent data or you are referring to a figure or similar – you should always spell out the numbers. Likewise, do not use “1st, 2nd, and 3rd”, but instead “first, second, and third”. Make sure to distinguish assure, ensure, and insure; all three mean to guarantee or promise something is true.
  • “Assure” is what you’re doing when you tell someone something is true.
  • “Insure” is when you back the guarantee, for example financially.
  • “Ensure” is when you’re actually doing something to make sure its true.
• Watch out for British spellings vs. American. Most important is to keep it consistent and use one or the other. A common issue is add an ‘s’ on the end of words with “ward” in them is British. I.e., “toward” vs. “towards” or “forward” vs. “forwards”.
• “Farther” vs. “further”. Farther is for physical distance (“A walked farther than B”), while further is for metaphorical distance (“A took the project further than “B”).
• “Who” vs. “whom”. These words are distinct in the same was as “he” vs. “him”, which people tend to know by instinct. Try substituting “he” or “him” to figure out which one makes more sense. “Who walked into a bar?” -> “He walked into a bar.” -or- “She gave the package to whom?” -> “She gave the package to him.”

And a few other more general tips:

• Spellcheck.
• Make sure to italicize anything in Latin. For example “in vitro” or “E. coli”.
• Use active voice when possible. It is always more exciting to say “I did this…” as compared to “This was done…”
• Be careful when using the word significant; in scientific writing you should almost always limit its use to the context of statistics. For other uses, consider alternatives like “drastic”, “vast”, “large”, etc.
• Remember that the paper isn’t doing the research, you are. Do not use lines like “This paper tests whether…” or “This paper explores…”. Ideally you should minimize referring to the paper itself, but at least use lines like “In this paper, we describe…”
• Make sure to keep consistent tense unless you’re making an active point of changing tense. When you have an option, present tense makes your work sound more active.
• Avoid informal constructs like contractions or words like “though” (use “although”) in any professional writing.
• Keep a list of common typos that you make that are not typically caught by spellcheck. For example “were” instead of “where”. Make sure to refer back to this list.
• Be careful about overusing words or expressions. Repetitions can make your text sound awkward.
• Avoid weak terms like “I hope to extend this work by…”. Always be more assertive: “I PLAN to extend this work by…”. Similarly, watch out for “may”, “could”, “attempt”, etc.
• In applying the rules above, make good use of text searches. They should enable you to find trouble spots.
• Did I mention that you should spellcheck?
• It is usually worth going sentence-by-sentence through your work and make sure that each sentence is contributing to your overall goal in an efficient way. Neighboring sentences with a similar meaning can often be combined into a single sentence. It is often even worth doing the same thing word-by-word. Most notably, people tend to overuse adjectives and adverbs (see my comment on “very” above). You might be surprised at how many words or sentences turn out to be superfluous. This technique is especially important for high profile papers where the text needs to be as tight as possible.

Research terminology

When introducing new terminology in a paper, it is always important to define it so that your reader will understand what you are talking about. We use many of these terms so frequently in Devolab that it can be hard to remember that not everyone will be familiar with them. Commonly used term to be careful with include: “EQU”, “Instruction”, and “Update”. We need to carefully choose the terms we use when talking about Avida, and as much as possible we need to seek consistency and clarity. Here are details about some of the more problematic or nuanced terms:

Also, remember the power of terminology. If a term doesn’t exist for a concept that you’re working with, you should seriously consider making one up. Be very careful that your new terminology is CORRECT and doesn’t CONFLICT with other terms. But such new definitions can not only make your paper more readable, but also allow other to build off of it more easily.

CPU Cycle

The unit of energy in Avida. Do not use “CPU time”; it’s odd to consider time as a unit of energy. Some papers have started to use “SIP”s which stats for “Single Instruction Processed”, which is a good alternative.

Fitness

The realized replication potential of an individual. In many Avida experiments this is the same as “replication rate”. Make sure to be clear that there is no explicit fitness function in Avida since they are self-replicators.

Generation Length

The number of CPU cycles it takes for an organisms to produce an offspring. Since the number of CPU cycles per update can vary based on the other individuals in a population, its not really proper to call this “generation time”. Until recently, this metric was called “gestation length”, “gestation cost, or “gestation time”, but this implies that it tracks only the creation of the offspring, not life up to that point. In Avida-ED, this is now Offspring Cost.

Metabolic Rate

Until recently, we had been using “Merit” for this, but metabolic rate is much more intuitive to the reader and makes more sense all-around when using the new energy model. In Avida-ED, this is now Energy Acquisition Rate.

Natural Organism vs. Digital Organism

Do not use “Real Organism” because it implies that digital organisms are not real. Likewise never refer to an Avidian as a “Simulated Organism”. Another good alternative is “biological organism”, particular when talking about ones evolved in laboratories, which are not technically natural settings.

Offspring

Not “child”, which is a human-specific term. Occasionally “daughter cell” is used for asexual organisms, but this should be avoided as well when possible.

Organism

Not “creature” or (shudder) “critter”. Also remember that they are not “people” either – it’s much too easy to anthropomorphize, which we need to avoid.

Runs

The most general way of referring to individual repeated runs of a given treatment/condition each with a different random seed, for the purposes of establishing statistical trends. Replicates is another option, but is sometimes an awkward word to use because we also talk about replication of organisms, but sometimes can work. Populations is commonly used in EC but technically incorrect if there is any sort of ecology going on.

Test Environment

Where we test phenotypic traits produced by a genome in such a way that the tests do not directly affect the ongoing experiment (though we may intentionally alter the experiment based on the results of the test). This is a better term than “Test CPU”, which used to be used but was not as clear. Often, this is referred to as an “isolated test environment” for extra clarity.

Treatment

A given setup or configuration in which a series of replicate jobs are run. Experimental configuration and environment are other options that are often not horribly confusing.

Unstable Genotype

These are genotypes that can produce an offspring, but even with all mutations rates turned to zero that offspring will not be identical to the parent due to an error in the copying algorithm. These differences are often referred to as implicit mutations, but people often have trouble understanding that concept.

Writing about digital evolution research

Getting started on the writing process

Every paper should address a well-defined scientific question and tell a good story in the process. Don’t expect the paper to fully answer the question, but it should at least shed some real light on it. The nature and background of that question will help shape the story you want to tell. There was a reason that you were excited about this research in order to start it, and that reason can also be used as a seed to grow the story from. Once you have the story in mind, write the abstract based on this story, clearly defining the question and briefly summarizing your motivations and the results. If you haven’t finished the experiments yet, that’s okay – just guess! You will update the abstract several times before you submit, but for now you basically have a blueprint for the rest of your paper, and an initial idea of the data you’ll need to include in it.

To better organize your thoughts on the required data, you should start planning the figures you want to include in the paper (see the next section for more on that). As the data comes in, those figures can be built for real. The abstract the the figures give you a proper skeleton for the paper, waiting to be fleshed out. Often a reader will scan the abstract and figures in deciding if it is worth making more of an investment in the paper, so you should make sure to take these seriously.

If you get hit with serious writer’s block, one trick you can use is to simply write an e-mail to a friend explaining the work that you’re doing. This is typically much easier than sitting down to formally write the paper and yet it provides a useful foundation to build off of. Once you have something in place, no matter how informal it may be, it becomes easier to develop it to a more polished form one piece at a time.

Titles

Most titles fall into three main categories:

• Stating the main result or contribution of the paper
• Stating the question of the paper
• Stating the topic of the paper

I tend to think that’s also the order by which they’re most effective, but that’s an overly broad generalization on my part.

So some title structures could be:

1. Selective pressure X promotes the evolution of Y.
2. How do we promote the evolution of exciting topic Y?
3. Computational studies on Y (or X).

There are a lot of different variants and pitches for each of these.

Per word, the title is the most important part of your paper. It determines (for the VAST majority of your readers) whether they are going to bother looking at it any further. The word choice also influences how prominently it will appear in different types of searches.

If you want your paper cited (and we all do, of course), title-style 1 is the most likely to get that done. People who don’t bother reading the paper will assume that they get the message from your title and use it to argue their points.

Authorship and Acknowledgments

Many people play roles in the creation of a paper, ranging from driving each phase of the process to just taking part in a casual conversation about it that helped refine an idea.

Who should be an author?

Authorship on a paper should only be given to those who played a significant role. There are three types of contributions to the creation of a paper: Intellectual, which involves generating and refining core ideas and interpreting data; Technical, including writing code, performing experiments and collecting data; and Written for figuring out how to present the results and writing them up. For those who contributed to the intellectual aspects of the paper, make sure to include as author anyone who’s input was essential to the core ideas of the paper. For those involved in the technical side, include only those people who contributed many long hours toward making this specific work possible. Anyone who is going to be an author on the paper should be involved in the writing, though it is possible for others to come in on that stage if they produce significant text or clarify it dramatically. A simple editing of the paper deserves and acknowledgment, but not authorship.

If you are still unsure if someone should be an author, it is generally a good idea to err on the side of more authors as long as each person’s contribution can be justified. Note that if ideas or experimental data were already published in a previous paper, that alone should not be enough to qualify someone for authorship, but that previous paper must be properly cited.

Who should be acknowledged?

This question is a bit easier to answer: whoever else was involved in any form. When in doubt, it rarely hurts to include an acknowledgment. If you feel like you were given a lot of useful advice at a group meeting, a thank you to the group as a whole will usually do. You also need to make sure to acknowledge the funding agencies and associated grants or fellowships that supported the work leading up to this publication; when in doubt ask your advisor.

What are the rights and responsibilities of an author?

All authors should read and approve a manuscript before it is submitted. As a lead author, do not submit a manuscript without getting permission from each co-author. Likewise, do not give permission to submit a paper without at least skimming it (and ideally more). I understand the temptation to improve your CV, but if the paper gets published, your name will be associated with it forever. Even if it does not get published, the reviewers and editor will still see it and link you to the ideas and quality found in the paper.

How to make a good figure

Figures are critical in any paper to express complex data and as a vital part of the storytelling. Your figures can make or break a paper.

There are several issues to keep in mind:

• Think about what information you want your figure to convey and what concept it should advance in your paper. Explore the variety of ways that you can get this across. There are lots of standard techniques (basic line graphs, scatter plots, bar graphs, pie charts, etc), but don’t dismiss the option of just creating a simple chart with the data you need. Likewise, don’t ignore the possibility of coming up with your own technique for displaying the data.
• Don’t forget that color can be a useful way of conveying information, but likewise you want to make sure your graphs are also readable by someone who is colorblind.
• Keep the figures as simple as possible where they still convey the message you want. Don’t include any extra information that doesn’t actually progress the message of your paper.
• Make an informative caption. Someone who has only skimmed the paper should be able to get the gist of a figure by reading the caption. The caption should spell out what data are depicted, what each line or color represents, and it should put the figure into some basic context.
• Don’t make your figures too small. You want to make sure your reader can see them clearly without straining their eyes. Likewise, make sure all labels are large enough to clearly read.
• When you design your figures, make sure you know the rules for your target journal: how many figures do they allow? Can you use color? Does color cost extra? Do specific fonts need to be used for the text? Are there limits on the size of the legend?

What to do before circulating a draft

You want to make editing a draft as easy as possible for your collaborators and you don’t want to waste their time fixing typos. Editing should focus on substance.

• Make sure you’re following the tips above.
• Spellcheck (in case you missed it before).
• Setup the page to be double-spaced (unless its a near-final draft).
• Make the figures full page (again, unless a near-final draft).
• Put line-numbers in the margins so that its easy for edits to refer to specific lines.

Tips on paper submission

Submitting a paper properly is as important as doing good science in terms of getting it published. Here are a few suggestions:

• Choosing your target journal can be a bit tricky. Consider topic, impact factor, turn-around time, track record of publishing digital evolution research, publication fees, and if the journal is open-access (you want as many people to see your paper as possible). Also important is acceptance rate, but that can be a bit harder to find.
• Make sure you read over the journal’s formatting guidelines carefully. You don’t want your paper rejected because your figures are in the wrong place or your citations are in the wrong format.
• Your cover letter is important. This document is the first item an editor will read and can sway their opinion on publication before they ever look at the actual article.
• Always include a list of suggested reviewers. Editors have a lot of trouble coming up with reviewers for your papers and as such appreciate a little bit of help from you in making some suggestions; don’t miss this opportunity. On the other hand, make sure the suggested reviewers don’t have a conflict of interest with you or your co-authors. Standard conflicts include co-authored papers in the last several years, joint grants, or a student-mentor relationship (in either direction).
• Don’t be too upset if your paper is rejected. Most of them are and editors are under so much pressure that they will often reject a paper without giving it a real chance, and there’s nothing you can do about that. If you are lucky enough to get reviews on the paper take them to heart, even when (or especially when) the result is a rejection. You may not always agree with the reviewers, but if they have a problem with part of your paper, chances are others will as well. Turn that paper around into a new submission, don’t give up, and you’ll eventually get it in print.

Responding to reviewers

(coming soon)

Post-publication

Once you have a paper published, you want to make sure to advertise the paper, both within the group (send it to the devolab mailing list!) and outside the group (certainly to the Digital Evolution Yahoo group at least). You also want to make it publicly available, if possible, so that people can look it up. Finally, if your publication is in a high-profile outlet, we should also talk about submitting a press release. These have shown a strong track-record of bringing press attention to the paper and its authors.