In May 2004 I completed a part-time PhD with the Open University. My research subject was the use of programmable logic in safety-critical systems. It took me 5.5 years from registering to submission, and 6 months more to viva (thesis defense for our trans-Atlantic cousins), do minor corrections and have the thesis approved. It developed from the work I did during a year of an abortive full-time DPhil. This article is intended to provide a list of the lessons I learned in these six years with varying degrees of pain. If you are at all considering a part-time PhD, read this. Some of it may prove relevant for a full-time PhD too. I'm writing from a British perspective. I've tried to explain terminology which may differ from overseas. If there's something you don't understand, please ask me. If it makes you feel better, there are quite a few of these items that I didn't do at the time, even though I knew I should... General guidance Don't do a part time PhD. I mean it. Whatever you gain in terms of job prospects, you will lose ten times over with the vast amount of time you'll sink into it and the frustration you'll have doing it.
Don't get married while you're doing the PhD. Ditto for buying a house. The PhD will take up more time than you can imagine. This is doubly true for the writing up phase.
Supervision Without a really good supervisor you have no chance of finishing. None. Meet your supervisor before you agree to start and ensure that he or she is a decent person who is technically able and who really wants you to succeed. Listen to your supervisor. <em>Especially</em> when they tell you that you're wrong or going off track. Do not lie to your supervisor. Ever. You can creatively omit to mention stuff to your supervisor, but on your head be it. Whenever you meet with your supervisor, have a clear agenda. Write up minutes afterwards. Check last meeting's minutes before going to the next meeting. Writing Stuff Agree delivery dates with your supervisor. Do not let them slip. Not even by a day. Keep the marked-up copies filed. Every comment is there for a reason. If your supervisor has misunderstood something you've written, it's not his fault for being technically crap, it's yours for not explaining it well enough. You can be accurate without being correct. Read Strunk and White, or a similarly definitive book on technical writing style and grammar. A key part of a PhD is learning to write well. Understand this and devote time to it.
Write completely unrelated stuff (fiction, magazine articles, letters) to improve your general writing style and grammar. I wrote Star Trek: Voyager fan fiction. It helped. Second Supervisors The Open Uni gives you several supervisors. As well as my main supervisor (Jon Hall) I had an external supervisor (Andy Vickers at my old software firm Praxis), a "research monitor" (recent Ph.D, Shailey Minocha) and a secondary supervisor (Darryl Ince). This is unusual, but not unprecedented. Ensure that you understand what each supervisor is for, and that they know that you know. Use their strengths. Start saving up now for the very large quantities of beer that you will owe them after you viva successfully. The Thesis Ensure that you understand exactly the form in which you must present your thesis to the examiners. Get used to using that form and page layout. Know your min and max word count and page limits. Keep an eye on your work-in-progress stats relative to them. Ask your supervisor for pointers to PhD theses which he regards as well-written. Read them. Ensure that you understand <em>why and how</em> they were written well. Make notes. Make sure that you have a firm thesis structure in terms of chapter titles by the time you hand in your literature search. Writing Articles Once you start making technical progress, start to look for conference calls for papers or suitable journals. <li>Agree a target conference and subject with your supervisor. Agree delivery dates both ways i.e. dates you'll send him versions and dates that he'll deliver back comments. <li>Carefully read all the guidance for authors they give. <li>Read previous proceedings of the conference or issues of the journal. Understand what sort of articles get published. <li>Submit on time, or a day early if you can in case something goes wrong. <li>While you wait for the decision, forget about the paper entirely. You won't be able to, but it's worth a try. <li>Please don't kill yourself when it's rejected. <li>Whether you succeed or not, read the review comments and pay attention to them. </ol> <a name="confs"> <h2>Conferences</h2></a> <ol> <li>Get along to at least a couple of conferences in your field during your time as a proto-PhD. The experience you get there is invaluable. <li>Take business cards. This is doubly important if there will be a substantial Japanese presence there. Ensure you understand card exchange protocol. <li>When you go out drinking with Canadian and Scottish engineers, understand what you're getting into. <li>Have a 20-30 second summary of your research area memorised. <li>Pick your conference tracks carefully if they have them. <li>Consider taking notes on each presentation you attend. <br>I took notes on my palmtop, converted them to HTML and put them on the Web (<a href="../FPL/index.html">here</a>). <li>If the conference is somewhere nice, organise a holiday around it. </ol> <a name="refs"> <h2>References</h2></a> <ol> <li>Read early. Read often. Read around the subject. <li>Identify the key conferences and journals, especially the ones you'll be publishing to. Go and read back as far as you can. Considering buying the entire proceedings for recent years. <li>From good survey papers, work through the references at the end and try to read as many of the significant ones as possible. Repeat, in a breadth-first search approach. <li>Get a PDA or small laptop and write up notes directly onto this in the library. <li>Write up your literature search as a chapter and hand it in to your supervisor for criticism. Many universities actually require this as a condition for transfer to full PhD student status. </ol> Bibliography Work out your bibliography scheme before you start. Keep a bunch of folders to store photocopied papers and articles. Label each one clearly with its bibliography label, ideally the same one that you'll use in your thesis to identify it. I use [name][year] and file them by year. So <kbd>moisset01</kbd> refers to Pablo Moisset's paper, published in 2001. I can then do an electronic search of my bibliography database for that reference, which tells me that I'll find it in the FPGA'01 conference proceedings pp125-133. The more data you have in your bibliography database, the better. A short note on the key point of the article would be really useful. Use keywords if you're disciplined enough. Now that disc space is cheap, try to find a PDF of the paper and save it in a suitable directory under the same name as your bibliography label for that reference. This makes it really easy to see whether you have easy access to a reference and saves on paper.
Technical Production If you are writing your thesis in the area of anything even vaguely technical then <strong>for fsck's sake stay away from Microsoft Word.</strong> MS Word is sold as a "word processor". Ever seen what a "food processor" does to food? Right. Use LaTeX, or something equally good. Word will make you sink unlimited time into just getting the text to look about right. Ergonomics Get a PC of your own. Take time and money to ensure that you are comfortable working on it for hours at a time. Get a good monitor. Saving a few bucks on a smaller or lower-spec monitor doesn't help you.
Follow the health and safety advice about not sitting in front of the monitor for too long at a time. Get hold of a gel wrist rest and try it out. I swear by them. Backups Backup. Weekly. At least weekly. Regularly verify old backups to ensure that you can still read from them. If you don't verify a backup, all it's doing is taking up space on a disk. Don't compress before backups. Plan on files getting corrupted; if a compressed file is corrupted, you're sunk. Backup to read-only media. A CD burner used to be perfect. Label your backups with the date and what they contain.
Use the Cloud (iCloud, Google Drive) but don't rely on it.
Move some backups offsite. I took a selection of mine to my parents' house when I go to visit them. If my house burned down, I could then still restore my work. Nowadays, the Cloud has mostly addressed this problem. Keep text files under version control where you can. I started off using RCS, but (in the last year) moved to CVS. This was great - it meant I could work on my laptop or on my desktop, and just commit my changes periodically without worrying about overwriting old changes. I then just backed up my CVS repository. Nowadays you probably want git or Mercurial. If you're keeping PDF files of references, backup these too. Bytes are cheap. LaTeX I used <kbd>pdflatex</kbd> to compile my TeX source directly into PDF. I drew diagrams with <kbd>xfig</kbd> and used a Makefile and <kbd>fig2dev</kbd> to convert the <kbd>.fig</kbd> files to PDF automatically. Have a <kbd>macros.tex</kbd> file with all your newcommands and newenvironments in one place. Use macros for everything. It is <em>much</em> easier to change a macro definition than hundreds of instances of it. Consider defining standard macros for references to figures, sections, chapters etc. Devise a standard label naming scheme and stick to it. I used <kbd>chap:XXX</kbd> for chapters, <kbd>sec:XXX:YYY</kbd> for sections in chapter name XXX and <kbd>fig:ZZZ</kbd> for figures. Have one <kbd>refs.bib</kbd> file for your references. Link to it for all your theses, papers and articles. Have a copy of Lamport's "LaTeX: A Document Preparation System" by your PC at all times. Put very little text in your top-level <kbd>.tex</kbd> file; let it include other <kbd>.tex</kbd> files. That way, if you need to produce your thesis in alternative formats (e.g. doublespaced, single sided, wide margins etc.) then you can make a copy of this file and change its settings appropriately. Miscellaneous Technical Stuff Do not upgrade your PC's OS without a <em>phenomenally</em> good reason. If you do, make sure that you can roll back to the previous version if it all goes wrong. I found I could use TeX, emacs, CVS and xfig perfectly well on both Linux and OS X. In the end, I preferred using the OS X laptop. Life Try to keep one. A balance needs to be struck. Tell your beloved ones what you're doing and why. Tell them that there will often be times when they have to come second to the thesis. Listen to what they say. Even in writeup hell, have regular time off writing. Keep a sense of proportion. Brief a good friend to keep an eye on you, and to slap you silly if you show signs of losing this sense. Getting married can torpedo your Ph.D., even at a late stage. If you do get married, pick a wife or husband who will insist on you finishing the Ph.D. and if necessary lock you in your office and refuse to feed you until you've produced 5 new pages in an evening.
The Examiner About a year before planning to submit, start looking around your conferences and areas of publication for an examiner. Pick someone established, not a young Turk out to prove themselves by eating students. There is a trade-off between the reputation of your main examiner and the ease of the viva. I went for a very eminent and well-respected professor with extensive industrial experience, and got well grilled as a result. That was what I wanted, but may not be what you want. Check your research school requirements for the examiner. You may find (as I did) that they must have done 5 vivas already before being allowed to be the main examiner. The Viva (Defense) Ignore your thesis until 2 weeks before the viva. Keep up with reading proceeds of the major conferences and journals though. Discover the day before your viva that you are expecting your first child. It's a great distraction and removes all viva-induced stress. But I realise this is quite hard to arrange... Prepare a 10-20 minute short talk at the start to summarise your work to the examiners. It's a good focusing activity and can clear up examiner misconceptions about your work. One week before viva I went through my thesis and wrote a 1-sentence summary of each page, generating an 8-page summary document. This was an excellent way of finding stuff quickly if I was asked about it in the viva. In the event I didn't need it - because everything was fresh in my mind as a result. When you're asked a question, pause (to make sure you've heard all the question), formulate a response and deliver it precisely. Make sure you answer the question you were asked, not the one you were expecting. You are not graded on speed of response. You will almost certainly be asked a "ground opening beneath one" question that makes you think your thesis is sunk. It isn't. If there were such a great hole in it, your supervisor or conference referees are very likely to have spotted it. The answer is in your thesis and it's OK to look. It may well be that either your examiner hasn't understood that point exactly, knows you've answered it but is testing you, or knows that the theory is sound but you've not explained a point explicitly and wants to hear you verbalise it. Corrections Your Ph.D. is not over until your corrections are done and accepted. So do them as soon as you can. But, if you can, wait for the official list to be sent to you so you're not correcting unnecessarily. Ensure that you have a copy of the source of your viva'd thesis so you know where you're working from. If you're using CVS revision control then "cvs tag" is your friend. Once you've sent off corrections for checking, start phoning around printers and binders so that you know what to do when the thesis is approved. Concluding Thoughts Good luck. You'll need it. I'm wide open to suggestions for other advice. Write me!