On PhDs

My academic journey was a turbulent one, replete with drama and politics. If I went back in time, I would probably decide against going down that path. But hey, I got through eventually!

I did my PhD in a technical field, so some of the below points are biased towards that. In a different context, some points are even worse (e.g. it's probably harder to find work as a History graduate than as a Computer Science graduate, across the board).

Should you do a PhD?

Take a step back really thing about why you'd want to subject yourself to that.

I want to better my job prospects

Search how much PhD graduates actually make on average compared to Masters graduates and most sources will tell you that they make less. Granted, this could also be because many stay in academia, which has lower salaries anyway, but there are many other reasons why you could make less.

Research by it's nature is not necessarily practical. Your job prospects are better with years of experience in industry, instead of the same years in research. Many PhD skills are not transferable to industry (and in fact many are actually bad practice in industry) especially compared to... actually practising those skills in industry.

Consider that you're also going to probably be using the best years of your life on this pursuit. Your friends and acquaintances will all be getting mortgages and starting families while you're slaving away to write papers for someone under subsistence wages. When I started my PhD (Jan 2016) London PhD salaries were £16k per annum. This is well below London living wage and nobody can survive on this without also working a normal job at the same time, which is difficult to juggle.

You might only really be in a position to start saving many years after your peers, putting you at a disadvantage as time goes on because of the compounding effects of that. At best, you will consistently be years behind peers of a similar age, and for what? The title of Doctor.

I want to learn / max out my stats

You would only be maxing out your perceived stats. The things I've learned during my PhD, I've taught myself. There is nothing I've learned during my PhD that I couldn't have learned in a different environment (including research in industry as opposed to academia). If you don't have the drive to teach yourself, the PhD environment won't magically give it to you. If you do, you don't need the PhD environment. In fact, the PhD environment will lead you to waste time on things that won't actually help you learn, due to perverse incentives, such as the "publish or perish" mentality.

Ask yourself if the title of Doctor is really worth the prestige. Other PhDs will tell you that there are some absolute blockheads that have achieved the same milestone, and similarly, I have met certified geniuses that don't even have an undergraduate degree. Ask yourself if the kind of people that would be impressed by your title are really the kind of people you want to be impressing.

It seems like fun

It's just not, believe me, unless you have a masochistic streak. And if you don't believe me, check the stats for mental illness and suicide rates in PhD students compared to the general population. Surely that can't all be correlation?

I'm being pressured by a parent or parents

Would they be more disappointed in you if you didn't go for a PhD, or if you went for a PhD and then washed out? If it's the latter, find a way to have a frank discussion with them about what you want to do. If you don't, it's very likely that you'll wash out, because it's really hard if you're not in it for you. You don't want to grow to resent your parent because of this. Blame me if you'd like; tell them that the evil Yousef convinced you not to do it.

The stats don't lie — if they want what's best for you, show them that what's actually objectively best for you is to not do a PhD.

I want to pursue a career in academia

This is the only valid reason to pursue a PhD, in my opinion, as a PhD is usually a prerequisite for anything else. Even then, consider carefully the nature of your PhD and the people you will work with, before taking the plunge.

Why was my PhD journey dramatic?

There were a lot of reasons why my journey was particularly rocky and it started even before I actually started my PhD properly. I was offered a PhD studentship right after my Masters (2014), at the same university I did my Masters. The supervisor was a great guy but I couldn't accept the position for a number of reasons:

So I went back to Germany for a bit and then struggled to find a position. Eventually I joined ‪Philip Torr‬'s research group, Torr Vision Group (TVG) at the University of Oxford, and was doing some Computer Vision research, which I do find interesting, though I struggle with some of the more maths-heavy areas.

TVG was quite exhausting, but I was in my early 20s and had a lot of energy. I worked quite hard and overdid it somewhat, as staring at screens for so long did cause me problems with my eyes, which led to four eyelid surgeries, but that's a separate matter. Philip was more than happy to supervise me for a PhD, but I needed to find funding, and I needed to find it fast, as I wasn't living in a sustainable way. My accommodation was free, as I got a room in one of the colleges, and as long as I collected receipts I could expense the food I eat, but that was about it.

I didn't have a lot of luck with securing funding and had some awful interviews. Philip said I could stay with the group and try again next year but I knew that couldn't work. So I looked for other PhD positions and found one in London that I thought was interesting. Philip was top of his field, at a top university, and wrote me a golden ticket reference, so I could get in anywhere. And so I went to London.

I joined Hamed Haddadi on the Databox project, a £1.5M EPSRC project, although my PhD funding didn't come from that, but from a joint programme between Queen Mary University of London and the University of Genoa. This would later become a problem, but let's not get ahead of ourselves.

Say what you want about me as a researcher, but if there's one thing I've proven again and again, it's that at least I'm an excellent engineer (not to toot my own horn). So right off the bat I did what I do best and built. The Databox was just a concept when I joined the team, and in order to do any research, I had to first make it exist. So I did. Eventually, others would build on the code I wrote, but for the most part it was down to me to implement things.

Over the following years, I worked hard. I lived in London, Genoa, Stuttgart, and travelled to a lot of conferences to present published work. I never found the technical part challenging (just as I hadn't found it challenging as an undergraduate) but the difference here was everything else was challenging. Circumstances had me pulled in 100 different directions and I had an abnormally large number of supervisors. This led me to produce a long, unfocused thesis, that I ultimately scrapped in its entirety. You read that correctly; I found myself at the end of my PhD with nothing concrete to show for it, through no fault of my own.

I dug up a draft the old thesis. You'll notice it's not only longer than the actual thesis for which I was awarded my degree, but also that it has almost 0 overlap. My old thesis is titled "Trade-offs in Edge Computing" and it's a mishmash of all the work I've done without making any tangible point. Hopefully this also explains why my final thesis is so unrelated with my list of publications. It's because I spent my entire PhD working on completely different things!

Anyway, where's all the drama you ask? There were a lot of problems, many of which I probably don't even remember that well, and many that were probably unrelated to the PhD and more related to living in London below the poverty line. Towards the end of my PhD, having basically restarted from scratch, I applied for an extension (unpaid of course). A silver lining when it comes to no longer being paid is that you well and truly no longer owe anyone anything. A failed PhD reflects badly on the university and can put supervisors in hot water when it comes to funding (deservedly!) so when faced with that, or giving you the freedom to do what you think is best, the latter is a much better option.

At any rate, I got the extension. For the application to extend, I had to write an explanation as to why I'm requesting one. At that point, I was approaching the end of my 4th year (funding only covers 3 years). To give you a truer insight about what was going on, I'm going to paste that explanation below, verbatim. Bear in mind however that I had to pick my words very carefully here and gloss over certain things. Just enough detail to get an extension, but skipping over the things that would rock the boat too much.

Outside of what you read below, I remember legal threats, nasty arguments, anxiety medication, dramatic weight loss. I remember living in squalor in many different countries, getting scammed at every turn, having my favourite jacket stolen. I remember the expensive London shoe box flat with an awful landlord and mould problem, as well as the 2 hour commute in Genoa which includes a long uphill trek (you really need a vespa to survive there btw) only to remotely connect to my London workstation (this was before covid, so people weren't accepting of the notion of remote work).

I remember a supervisor (or maybe he wasn't, I've lost track) telling me after a particularly spicy Skype call I had with another supervisor overseas, that he's impressed with my restraint, as he would never let anyone talk with him like that. I suppose dealing with contentious situations is one thing the PhD inadvertently taught me?

Fundamentally, a string of unfortunate circumstances, including academic politics and multiple changes of supervisor, have led to my research topic being almost completely changed, and a lot of work that I have done to be disregarded, leading to the significant delays in deepening other work. By the time I started my writing up period, I was left with many disjoint units of work.

Before I summarise what led to this, I would like to just point out that during my second year, I took a 3-month internship position at Bell Labs Stuttgart under the recommendation of my original supervisor. I did not interrupt my studies to do this, as the research I was doing there was very related to my PhD, and the rules stated that “during your period of interruption you must not work on your research”. My original supervisor recommended that I interrupt, but I didn’t want to have this work barred from my thesis. Since my topic changed after that, the work of those three months really was unrelated, and in hindsight I should have interrupted.

The first cause is a very unusual level of supervisor churn. When I first started my PhD, I had a primary supervisor and a secondary supervisor. Later, well into my PhD, I found out that my PhD was part of a collaboration with another university (Genoa), and that I had a supervisor there also. I am officially part of the CogSci research group, but I have not ever been involved with anyone else in that group, and have not to this day.

I started off by doing a lot of engineering work for my original QMUL supervisor, with the expectation that we will be able to create a lot of research output through the platform we build. This project was a collaboration between QMUL, Nottingham, and Cambridge. I was essentially the only person working on building this platform at the time, and I build the first prototypes, and ran some evaluations.

Towards the end of my first year, my supervisor got a large EPSRC grant for this project, so a lot of people started working on it, funded by that grant. I collaborated a lot with academics from the other universities, and I would go as far as saying that one academic from Cambridge gave me much more research advice than my original supervisor ever did.

During my second year, the team at Bell Labs took up that supervision mantle unofficially, and I continued to work closely with them even after I left. While I was there, my supervisor transferred to Imperial and took the project with him. He would continue to unofficially supervise me, and I would continue to work on this project (including with Cambridge etc). During this time, my new official QMUL supervisor was nominal.

As part of the collaboration with Genoa, I was supposed to spend a year in Italy. Eventually, I spent a few months there, where my research became increasingly strained. Even in Italy, I had multiple supervisors though. I had an official supervisor, but I did not report to him (I only actually ever spoke to him once). Instead, I was to report to the person below the person below him (two steps) in hierarchy.

My time there was cut short in part due to my research requiring my presence in London, and breakdowns in communication that began long before I went to Italy, creating the divergence in research in the first place. The general consensus at my stage reviews was that I needed to limit my scope, but by now I had expanded my scope even more in an attempt to satisfy the collaboration.

In London, I joined my original supervisor as a visiting student at Imperial and was there for a large part of my 3rd year. This caused a lot of problems at QMUL, which ultimately led to ties with my original supervisor and the project being severed. These problems included the collaboration with Genoa, which eventually led to my supervision there being changed too.

Throughout this, you must imagine the logistical overhead of relocation. I had also moved to West London as a visiting student at Imperial, and had to move again when that ended, since it was very difficult to commute to QMUL. At the time, the CS building was being refurbished and our offices demolished, so it made sense for me to simply move my work to my office at Imperial.

Anyway, back at QMUL, at this point, I tried to focus my research, and take the work that I had done with Bell Labs further. My nominal replacement QMUL supervisor, who was involved in the Genoa issues, changed. My two new supervisors (who are my current supervisors) advised me together with the team at Bell Labs.

I soon realised that the research I’m doing with Bell Labs was stagnating, and incredibly forced, since I tried to shoehorn my original research into it in order to limit the scope of my PhD. After discussing this with my current supervisor, we decided that there was no point in wasting time with pushing on research that will not get anywhere.

Finally, we come to the change of direction. By now I’m on the tail end of my PhD, and I push hard to tease out novel contributions. I only had enough time to get out one small publication post-change (the rest are from the old research) but the work was finally promising.

I enter my writing up phase with the expectation of threading together all these disjoint pieces of work into a cohesive narrative of sorts (even though it would be a stretch). Please also note that my current supervisors were not significantly involved in my original work, so I was really the only one with the complete big picture.

This merging proved to be very difficult to do. The consensus among my panel (the new QMUL supervisors, and my new Genoa supervisor) was that for my work to have enough weight come viva, I would need the focus to be much tighter.

I have therefore attached two versions of my thesis. The first is the one where I tried to include all the work I ever did. The second is the focused version, which I have been working on just these past few months.

I am now just over a month away from the end of my 4th year. I had never planned for it to take this long, but I hope that my account of the past events helped illustrate that this was out of my control.

I also understand that working a job is not enough a reason to extend a registration, however I could not have planned for this, and I have had no choice but to work part time in order to survive in this city on no other income. I try to limit my work to weekends however, with minor weekday tutoring gigs, such that I can work on my PhD "full-time".

All these factors delaying my work had the effect that I must repeat stages of work now that I would rather have already done last year. I have been pushing to get all this done, and at the moment, all I need to do is expand on the evaluations that I have already done in order to add weight to the research contributions I have made.

This unfortunately means repeating some simulations that I have already done under different conditions, and with better datasets, and creating more informative plots to illustrate the novelty. This is straightforward work but takes time. The results would bolster my second to last chapter, and hopefully give my thesis the depth it needs so that I can pass my viva comfortably. Attached is a plan for doing this work in the extra time requested.