Robert Ghrist
December 2014; updated, April 2017

Academics have been leery of self-publication of scholarly works. The trepidation is reasonable, given the desire for signaling quality and declaring that the work has been carefully vetted and proofread. I have been experimenting with self-publishing for several years. This is a record of my reasons, processes, and outcomes.


PART 1: print

In Fall 2014, I published a book, Elementary Applied Topology [“EAT”, below], aimed at graduate students and researchers in the mathematical sciences.  The book is dear to me and represents the synthesis of a significant portion of my professional career.  It is an idiosyncratic text, but not at all unpublishable by a top venue.  I chose to publish my book using Amazon Createspace, a print-on-demand service.


  1. I retain the copyright and am completely unrestricted with regards to the publication of the text. I choose to keep pdf copies of EAT on my website, available for free. There are a few publishers who will do that (Cambridge is quite good about it), but not many.

  2. I set the price of the book.  As long as it is higher than Amazon’s costs, I get the remainder of each sale as a royalty.  To experiment with their pricing structure, check out their on-line calculator.  The pricing is such that I can set a list price of $24 for EAT; Amazon tends to discount it to about $22 (subject to change! I don’t control their discounts…), and I still make a very fair $10 royalty from each sale. Nobody is going to get rich by selling mathematics monographs, but it’s nice to have a royalty higher than $2.

  3. I’m allowed to be different. I really wanted to put figures on the front and back cover of my text; a lot of publishers of monograph series in Mathematics have rigid style formats and won’t allow that. Also, I decided to put some Joycean puzzles hidden in the book (allusions, secret codes, hidden theories, etc.) to keep me sane while dealing with writer’s block. I didn’t have to ask permission from a publisher to do something weird.

  4. Because this is through Amazon, it’s easy for buyers to order the book: Amazon “Prime” shipping is in effect for prime members.  They take care of payments, returns, advertising, analytics, etc.  It works in the USA and most of Europe at present, with other (but not all!) countries eventually coming on board.

  5. The quality is very good.  They cannot handle hardbound books, but their soft-cover books appear to me to be of the same quality as other soft-cover math texts by leading publishers.  I chose to make the book black-and-white, but color is available for a larger price.  The cover is full-color.

  6. Updating the book is easy.  At any time, I can upload a new version of the text.  No permissions required; no delays. This makes it easier to publish a book in a cutting-edge subfield, since changes can be incorporated into updated editions.

  7. The distribution network is wide.  Depending on what distribution network one chooses (see below), the book can be made available for distribution by physical bookstores and for purchase by academic libraries (albeit with smaller royalties).

  8. Publication turnaround time is very fast.  You upload the text and the cover (see below), order a proof copy, then press the little red “PUBLISH IT” button.  Done.

I am very, very happy with my decision, and am convinced that academics should start using this platform for the publication of books.  

The objections that I can think of are these:

  1. But it’s not refereed”:  well, it can be, but you have to get your friends to do it, and it requires calling in some favors.  Let’s be honest.  Most math research texts are not carefully vetted and proofed.  Yes, the publisher will sit on the draft for months on end, but most referees are not going to do anything that a dedicated reader won’t.  Because one can change the text with ease, it’s perfectly reasonable to wait for readers to contribute error reports.  Publishing errata (say, on a web site) and updating the text every year seems to me to be a reasonable solution.  Of course, mathematicians hate to make mistakes, but let’s not pretend that it doesn’t happen.

  2. Publishing with XXXXX signifies quality”: no, it doesn’t.  Don’t make me link to books on quantum knots and consciousness, the mystical golden ratio in nature, fractals and markets, and all manner of pseudo-mathematical claptrap published by major presses. These are exceptions, to be sure.  Publishers in mathematics are desperate for new books – the only way to make money on texts with low sales is to make a lot of books – and this situation is only getting worse. If you think to yourself, “But all the books by XXXX Press that I own are really high quality,” then you are forgetting how natural selection works. Go to a catalogue of “New & Upcoming” to get the picture of quality at the margin, before evolution thins the herd.

  3. But how will my book get noticed if I’m not already famous?”  Yes, you won’t get any free marketing, though Amazon may promote its own books more highly – I suspect, but don’t know.  If you are trying to get established, here is some good news: most books in Mathematics are poorly written, poorly organized, and poorly illustrated.  Assuming that we’re all smart and doing good math (a decent assumption in this field), it’s not hard to beat the competition as far as making beautiful, readable books. Quality (and a good web site) helps a lot.

That’s the good news – you get a reconfigurable, you-own-it-all, pays-a-decent-royalty book backed by Amazon’s incredible distribution network.  What’s the bad news?  You have to learn a few things about book production. It’s not enough to be good at LaTex – you also need to know a few things about book design.  Here are a few tips.

  1. To publish on Createspace, you upload the interior of your book as a pdf.  You need to have it perfect before they will let you publish the book – no sloppy LaTex job allowed.

  2. You will want a style file that is more customized than the usual “book.sty” in latex.  It takes a lot of tweaking to get something that looks good but not generic.  Fonts that support LaTex math are hard to find, and most are not so good. For EAT, I used cmbright and am happy with the clean, geometric look of it.  It looks even better in print than on-screen. Make sure to use T1 font encoding, like so:


  3. Amazon's Createspace lets you specify the size of your book, and it’s very flexible.  However, you need to get your pdf margins perfect.  You also need to put in a gutter (look it up) and it takes a bit of work to get the gutter margins right.  (My only complaint about the print quality is that the margins are slightly inconsistent, a fact I noticed only because I had to be so picky before uploading…) Amazon's Createspace will also provide you with a (free) ISBN, should you choose to have them assign one.

  4. To crop the pdf, I use Adobe Acrobat Pro.  Highly recommended, for several reasons [see below].

  5. Be sure to embed your fonts: Adobe Acrobat will help with that.

  6. You need to make sure the pdf is PDF/X-1a compliant (X-3 should be ok as well). The usual pdflatex or dvipdf commands will not work, unless you’ve set your flags properly (which I never got right). My solution was to use dvips and then feed this to Adobe Distiller.  That program does a great job in setting up the pdf to be nearly 100% /X-1 compliant.

  7. Great, but not perfect, and you need perfect to pass Creatspace’s auditor.  When you upload the pdf to Creatspace, it will run automatic checks for you.  You may be disappointed to find hundreds of errors, but don’t worry. Most of them are about too-low resolutions on latexed formulae.  These are ignorable, and won’t impact the print quality at all. Their setup is not optimized for LaTex users. 

  8. However, if you get *any* errors on margins or figures, then you have more work to do. The best method I found for fixing these was to use Adobe Acrobat’s Preflight (CTR+SHFT+X) tool. Under PDF/X compliance tools, select Convert to PDF/X-1a (COATED FOGRA39). Clicking Analyze and Fix did the job for me.

  9. OK, that’s the interior of the book, but you’re not done.  You need to make a cover.  I thought this would take me a day or so of work. WRONG. Plan for several days.  Createspace will give you a downloadable template, but it’s on you to make the cover right. You need front, spine, and back on a single-page pdf. A little googling will get you a barcode generator for your ISBN. You won’t be able to LaTex this.  Time to break out some professional-grade drawing software and get to work.  I used Adobe Illustrator, but cheaper and more user-friendly software is out there.  Just make sure it is beautiful, since the cover is what sells the book. 

  10. Don’t use a generic font on your cover. I wound up having to make my own for the title of EAT. You don’t need to descend to that level of hell, but, please, don’t slap down a title with SliTex.

  11. When you upload everything and it passes inspection, then the hardest part begins: waiting for Createspace to send you a proof copy.  They are still separate from Amazon, with a separate shipping facility, so don’t expect to receive your book in 1-2 days: more like 4-5. However, once you approve and press the PUBLISH button, Amazon will have you listed and available for purchase within hours. Sales reports and rankings are compiled on an hourly basis, from what I can tell. Amazon gives global rankings of their books, and breaks it down along a subject tree. I know exactly how well my book sells as compared to the other topology books out there. That's useful.



PART 2: electronic

In 2015, I began working to write a new type of calculus textbook that would: (1) update the curriciulum to be better integrated with modern applications, especially to data; (2) be graphics-based but with text-based search features; (3) be fully electronic and look good on a cell phone or kindle device [with color, of course].

The result of this is a series of calculus comictexts, called CALCULUS BLUE , covering a full semester (and then some) of multivariable calculus, broken into four 400+ page volumes. At with EAT, I am publishing the BLUEs through Amazon, but this time through the Kindle store.

Amazon makes it ridiculously easy to do the publishing. After generating the pdf of the book, including a cover image, I use the free Kindle Textbook Creator to process and package for uploading to the Kindle store. This is the right way to go for fixed-layout, structured content, and the resulting text will not re-flow like most kindle books: the formatting is fixed. If you are, say, in the humanities and wish to publish a text-based book for the Kindle, it's better to use other tools, described here. Don't let the [Windows XP] look of the KTC software discourage you. It's really easy to use.

Amazon lets me set the price of these e-texts, with differential price based on country: my books costs much much less in Brasil and India than in the USA and UK. They automatically handle returns, royalties payments, currency conversions, and tax collection. Given that there are at least two dozen countries involved, that's a big help.

Downsides? One. The royalty plan is 70% if you publish exclusively with Amazon; otherwise it's 35%. But read the fine print: the 70% royalty plan comes with delivery fees on a per-magebyte basis. These fees are ridiculously high, so much so that it costs about $7 for one of my graphics-intensive texts (more than the cost of the text!). So, I am stuck with a 35% royalty; half of what Apple pays. Am I tempted to work with Apple? No: their distribution network is miniscule and their publishing tools are awful to work with unless you generate all the content from within their system.

The truly difficult part of e-text production is getting something that looks really good. Graphic design is hard and there are good reasons why you should hire a designer to work with you. I do my own art and design, but it's very difficult (for me). For example: font designs (3 families, six months), color schemes (ugh!), proper embedding of fonts, gradients, etc. [hello acrobat pro], and, most difficult of all, graphing surfaces, vector fields, and 3-d objects (which required learning and fusing several software packages). YMMV by field, but, unless you want to wrangle with copyright issues for others' images, you will need some help with artwork. Good luck!



I have not been a vocal participant in the anti-Elsevier or open-journal debates of the past several years. I have no particular animosity towards any publisher, but for the collective industry, I do predict hard times, well-earned. These remarks are less about how awful publishers are and more about how satifying it is to take advantage of Amazon's tools to distribute my content to the world.

I hope that other mathematicians and academics find these comments useful.