Reverse engineering your CV

I recently discovered a great podcast called The Startup Scientist, of which I have now swallowed up all episodes – thanks to Daniel Lakens for the recommendation! The podcast is about treating your academic career as a start-up. I have made this comparison before when giving talks at career orientations but wasn’t able to distill this idea, so I was excited to discover that Dan Quintana did exactly that, so now I can add on to it in this post.

The episode I want to write about is Reverse Engineering your Career Goals. Although I do often say that my career so far is a series of coincidences, I do think this type of reverse engineering has played a part in it.

Towards the second half of my PhD I had a lot of doubts about continuing in academia – I really wanted to, but I was aware that it would be very difficult to get a position. I wanted to do a postdoc, but only if I had a reasonable chance of getting a position afterwards. I was not sure how to estimate this chance, so I used the approach below.


1. Find representative data

I first studied the CVs of Dutch professors I knew, and concluded that most of them received independent funding a few years after their PhD. But these professors were already professors for a few years, so I decided their situations did not apply to me.

Through the funder’s website, I found a list of people in related disciplines who received this type of funding that year. Since I was a year away from finishing my PhD, and these people got their PhDs maximum 3 years ago, the gap between me and them narrowed.


2. Generalize

After searching for CVs of my “prototypes”, as a good machine learner I tried to find patterns. Although the people were all quite different, and I only had a few examples (I wasn’t doing this automatically, although that would have been an awesome side project), it was easy to spot several things they had in common.

I did this during my pre-Evernote era so sadly I do not have notes on what exactly I discovered, but I do remember two things in particular:

  • All recipients had an impressive (at least for me) h-index. The minimum I found was 6, but the values between 9 and 12 or so were more common. This will vary per field, but for context, the professors I thought were not representative, would be in the 20+ category.
  • Many recipients had an extra “thing” that was different from most others, like their own company, an organization they volunteered for, etc.

3. Multiple time points

Next to looking at people’s websites, it’s also helpful to search for “Name CV filetype:pdf”. Using this method, I was also able to find CVs of the same people, but from a few years ago. This had several benefits.

While the current CVs looked quite intimidating to me, the “time travel” CVs were much more relatable. Instead of learning only about the “value” of a CV, I was now learning about the “gradient”, which would be easier to apply to my own situation.

Of course, it is possible to do this for any CV, by removing all the things from the recent years. But the nice thing about the real “time travel” CVs was that I also saw how people changed the way they presented the same thing, that both CVs already had. For example, a side project that might have been insignificant in the “time travel” CV, was described in more detail in the current CV.

4. Predict?

Now that you’ve estimated your function where the inputs are the CVs at different time points and the output is receiving funding, you could try to fill in what’s missing from your future CV, to get the same output.

Don’t forget that there are multiple solutions (so different inputs will lead to the same output) and noise (so the same input might lead to a different output). Since I’m on a roll with this analogy – there are also lots of other inputs that you might not even be considering yet (life and stuff).

I did not actually create any concrete plans of what I was going to do, based on this informal study. But I suppose that the information got deposited somewhere, and helped me make choices that would later point me in the direction of a solution.

A few thoughts on mentors

As promised, a post on finding mentors and being a mentor. This is something I’ve been thinking a lot about recently, so it’s difficult to distill everything into a structured post – but here a few thoughts / tips that might be helpful.

1. You need more mentors than only your supervisor

Even if you have the best supervisor in the world, they will not know all the things you might need advice on, like time management, planning your career or putting your health before your project. This is even more true if your background is different from the supervisor’s, and they have never been in the situation you might find yourself in.

Therefore it’s important to find other mentors who are more similar to you. Gender is an obvious characteristic that comes to mind – here are studies showing that women assigned to women mentors are less likely to drop out of science and engineering, or more likely to become professors themselves.

I think this is similar to my experience. As I have written before, I had a great advisors during my PhD. I would have probably laughed if I had to join a mentoring program and meet with another professor, just because we both happened to be female. But I do think that meeting some amazing women mentors along the way, is what convinced me to give a career in academia a try.

That being said, there are other characteristics that can define what similar means for you, but are more hidden, such as being a first generation student or having a health issue.

2. Mentors can be anyone

A mentor doesn’t necessarily have to be a more senior person in the same field. If you see the word mentor as “a person you can learn something from”, it becomes easier to think about this. Here are some (perhaps less traditional) examples of who I consider mentors:

  • Assistant professors in different fields related to computer science. We have a Slack group where we share advice, our failures (#rants) and successes (#humblebrag).
  • Students who I do not supervise, but who inspire me because of how they approach their work and/or life, and/or who teach me how to be a better researcher and supervisor by sharing the experiences they have.
  • Academic community on Twitter, where lots of amazing advice is shared.

Note that most of these people probably do not consider themselves as mentors in these situations! These conversations do not start with “will you be my mentor?”, but with genuine questions about a particular topic, that the others might have more expertise in.


3. Be a mentor too

Since anybody can be a mentor, you can be a mentor too!

If you want to benefit from others’ mentorship, you have to be able to offer something in return. This might be difficult to imagine if your mentor is more senior, but they can probably still learn something from you too. If this still does not apply to you can pay it forward by mentoring other students.

In all cases, be a good mentee – here’s a recent thread on the subject:

Finally, I’d like to share this post on service as leadership – mentoring isn’t a chore you have to do, but an amazing opportunity you get to do.


7 things I wish I had done during my PhD

Every so often there are threads on Twitter about what people wish they would have would have known before starting their PhD, or would have done differently in retrospect. Here is a thread with lots of great advice by David Schoppik and another one by Jennifer Polk. I haven’t responded to either question, because there is so much to say that I can’t fit into 140 characters. However, I have already been keeping a “wishlist” of sorts, so I thought this was a good opportunity to finally turn them into a blog post. Here they are, the things I wish I had done during my PhD.

1. Having a lab journal

I somehow managed to miss out on this concept completely. Maybe I had heard about it, but dismissed the idea because I didn’t work in a lab. I only really found out about it when I was about to start my tenure track position, and was reading “At the Helm” in preparation.

Sure, I had a notebook. I would use it to make notes in meetings, draw toy datasets, write down tasks as they came up… anything, really! But none of these things were intended for anybody else, including the future me, to read.

In retrospect, it would have been helpful to have a central place to record ideas, different (failed) experiments, and where I ended up storing my data and code.

2. Having a todo list

This might be a surprise to many, but I didn’t have really have a todo list during my PhD. I would write down tasks as they would come up – for example “prepare presentation for lab meeting” – in my notebook. If I didn’t get a task done 2-3 pages later, I would copy it over to the current page I was on.

I don’t remember forgetting to do anything important and I didn’t miss any deadlines, which probably gave others (and myself) an impression that I was an organized person. But the 2017 me is overwhelmed by the idea of this “organization system”.

3. Spending more time with other PhD students

I don’t mean with this point that I didn’t spend any time with friends or colleagues. I did my PhD in in the same city as where I got my other degrees, so there were lots of friends around. And I was in a great lab, where we would often do social activities together, and would see each other as friends. I realize that I’m very lucky to be in this situation.

But most of friends were not doing PhDs, and with my colleagues, often it was more relaxing to talk about topics outside of work. So it was great to meet other PhD students, for example during courses, and share experiences about writing, teaching… anything that might have been challenging. I should have done that a lot more! Maybe I would have learned about “lab notebooks” and “todo lists” 🙂

4. Seeking out more mentors

As I wrote above, I was in a great (though perhaps small) lab. My supervisors were both inspiring scientists, and very kind people. But even despite these favorable circumstances, I didn’t always dare to tell them what was on my mind. How was I doing with my research? Was my CV maybe good enough to apply for this scholarship? Did I have good chances of getting an academic position? Questions I was too scared to ask, because I thought I would be laughed at, even though I logically knew that wouldn’t happen!

But things changed a bit when I did an internship, and met two very different mentors. They were closer to me in age and career step – both postdocs at the time – and were women. They saw right through my self-esteem issues, and made me a bit more confident that I wasn’t entirely delusional about my aspirations.

5. Applying for all the things

In the Netherlands, as a PhD researcher you are an employee, not a student. I had a salary and my travel expenses were reimbursed. Therefore I never felt the need to apply for any financial support.

As for awards, most of the time it either didn’t cross my mind I should apply, and if it did, my imposter syndrome didn’t let me. It didn’t help of course, that the one scholarship I really thought was a good fit (Anita Borg Memorial scholarship) was rejected three times in a row.

In retrospect, I think applying for more things would have made the applications I really wanted, like the Anita Borg one, a lot better. Not to mention the benefits for applying for larger grants later on.

6. Joining Twitter

Although I had an account for years, I didn’t start using it until half-way through my postdoc. Maybe a funny story is that this all happened because of a grant I applied for. The best submissions would be advertised via Twitter, so I thought I should at least see what people are saying about my submission (not much). But since I was now checking Twitter every day, I also started following more accounts, engaging in conversations etc – and never left.

Being part of the (academic) community on Twitter has been pretty awesome. From excellent advice about applying for jobs, to thoughtful threads about academic culture, to cat pictures (#academicswithcats), there’s always something to motivate me or cheer me up. Through Twitter I found many friends, role models, and from time to time, even people who were somehow inspired by me. I cannot stress how essential this has been in times of existential crises almost inevitably associated with being a postdoc.

7. Blogging

I had a blog on and off during my PhD (see My relationship with blogging), but I didn’t really dare to write anything, let alone tell other people that I have a blog. I had a blog, but I wasn’t blogging.

And that’s too bad. Because since I really started writing and sharing posts (although I still find I’m often outside my comfort zone), lots of interesting things happened. Next to improving my writing and getting me invitations to give talks, blogging has given me a bigger sense of purpose. Related to the Twitter point above, this has been essential for dealing with setbacks.

I hope these are useful whether you are doing a PhD or are already done – it’s not too late to start! If you have any other advice you’d like to share with others, please leave a comment below!

Firsts: Submitting and revising a journal paper

This post contains the history of my first journal paper, “Multiple Instance Learning with Bag Dissimilarities” (also available via my Research page).  Recently I shared some examples of cover letters and responses to reviewers with people on Twitter, so I thought it would be informative to put it all together in one place, including a timeline of the process and take-home messages. Note that this is not a guide how to write a paper or how to respond to reviewers – but if you are looking for that Dr. Raul Pachego-Vega has lots of resources for this.

On with the story of the paper. The files (draft, original and revised submissions, cover letter, reviewer response) are all  here (.zip), but the post should be readable without them.

Draft to first submission

In my PhD years 1 and 2, I had a few workshop publications which were exploring different aspects of one idea, and it was time to put these results together into a journal paper. I made a first attempt to organize all my ideas in November 2012. This first draft then went through several iterations of discussions and comments with my supervisors. Finally, on the 25th of March 2013 I submitted to TPAMI. I think the plan might have actually been to submit to Pattern Recognition. But as I understood at the time, TPAMI was more impressive to have, and had a faster review process, so it was worth a shot.

When submitting, I did not consider anything else I would need other than the paper, like a cover letter! Therefore my cover letter was very short and uninformative. I only mentioned that the journal paper was based on earlier conference submissions, but not what the differences were. It seemed obvious to me that the journal paper was so different, that I didn’t need to explain this. Of course, after a few days I received an email (I remember having this stomach sinking feeling) that I needed to provide a summary of changes, which I did. It seemed my paper was still under submission – crisis averted!

Take-away: if the paper is based on any conference publications, explain the differences in the cover letter, even if you also do this in the paper

Rejection and another submission

The part about the fast review process was true. On the 5th of June 2013 I received the decision that the contribution was not significant enough for TPAMI. Since my paper wasn’t immediately rejected after submission and actually went to reviewers, I was quite satisfied with this result.

After updating the paper according to the useful comments I could extract from the reviews, on the 26th of June 2013 I submitted the paper to Pattern Recognition. Once again, the submission system caught me by surprise! While I now had a better cover letter, I now also needed to provide a “graphical abstract” and “highlights”.

Take-away: go through the submission system of the intended journal before you actually want to submit there. Another surprise I’ve encountered in other journals is that I could suggest names of reviewers – it is good to think about this beforehand, while you are writing the paper.

Major revision and responding to reviewers

As expected, the review process at Pattern Recognition was a bit slower. On the 23rd December 2013 I received a “reject and resubmit” or “major revision” decision. This was a more hopeful situation than with TPAMI, so I started revising the paper and writing my response.

A useful structure for the response is:

  • reviewer comment (in bold)
  • your response
  • quote from the updated paper which shows your changes (in a different text color)

Take-away: simplify the life of the reviewer, they likely do not exactly remember your paper and do not want to go through the whole thing, switching between the document and the updated paper, to see if their proposed changes have been made.

What I did wrong the first time around, is that I would do the suggested change in the paper and in the response, then would discuss both the paper and response with my supervisors. This was not productive – since they proposed changes to my change, I would have to modify two files!

Take-away: write the response first, include proposed changes in the response, then discuss with supervisors, then add changes to the paper!

Another annoying thing was that in my response I was referring to section numbers and references in the original paper. But since these would get updated (due to new sections or references), I would have to keep changing these by hand in the response. But, it turns out there are LateX packages for this too! See this answer on Stackexchange.

Besides the responses, we added a “cover letter” to the beginning of the response, explaining that we prepared a revision and summarizing the changes made. After a month or so, I submitted the revised paper! I was confused about filling in a “cover letter” text field in the submission system – after all, I now had a whole response to reviewers, that was a cover letter in itself. But I think I just copy pasted “cover letter” from the response, with a comment that detailed responses can be found in response.pdf.

Minor revision and accept!

Then finally, a long-awaited email came on the 20th of June 2014 – “We would be happy to publish your manuscript […] in the journal provided that it is revised in accordance with the enclosed referee comments.”

A minor revision! I remember exactly where I was at the moment – on a camping in Sweden, getting ready to celebrate midsummer. I was walking back from the bathroom to our tent and decided to check my email on my phone, and there it was. I’m pretty sure I jumped and yelled “yes”, or something of the sort. But it was great that I was already at a party, so I could celebrate this event immediately 🙂

Take-away: don’t forget to celebrate!

I sent in the revised version a few days later, and on 21st of July 2014, the paper was accepted, and in early 2015, published. A nice detail is that at the moment the paper was published, it already had a couple of citations – because I uploaded a preprint to arXiV back in 2013. See my blog post about this and a great post by Niko Kriegeskorte if you are still unsure.

Take-away: upload your papers to arXiV

Share your experience

Have you had a very long, or perhaps a very short review process? Surprises you encountered during the submission process? Or do you have any other tips about submitting papers you could share? Please leave a comment below!

On being an employee vs a student during your PhD

This is probably my most retweeted tweet to date. Since this seemed surprising to many people, I thought I’d explain a bit more about what this means.  [Disclaimer: most numbers are estimates based on my sample size of 1 – if you have more detailed / up to date information, please leave a comment!]


First of all, being an employee means that you get a salary. As an example, let’s take a look at some PhD vacancies in the Netherlands. Many of these can be found on AcademicTransfer. Here are two positions in the group I’m currently working in.

At the bottom you will see “Gross monthly salaries are in accordance with the Collective Labour Agreement of the Dutch Universities (CAO NU), increasing from € 2191 per month initially, to € 2801 in the fourth year.

You might think the salary is there because it’s a STEM PhD, or because of the PI. Well, here are two positions in different fields and different universities – investigating why Dutch people are so tall and how people communicate positive emotions. Both mention exactly the same numbers! This is because of the labour agreement, which defines the salary, and a number of other benefits (more on this later).

[Note: at the time you are reading this, these exact vacancies might be closed, but you can find other similar positions on AcademicTransfer].


Of course, the salary doesn´t say much if you are from a place where the cost of living a different. First, there are taxes. The actual salary you get depends on a number of things, like your savings, whether you rent or own a house (yes, this is possible during your PhD) etc. To give an indication, in 2011 my 1st year PhD after-tax salary was around 1400 EUR, and in 2014 my 4th year PhD salary was around 1900 EUR per month.

The biggest cost is where you live. This varies between the Dutch cities, with Amsterdam being the most expensive one. Delft, where I did my PhD, is a bit on the expensive side as well, but it’s doable. Even if you don’t search too long for a great deal, you could rent a room in a shared apartment for starting at 400-500 EUR or so, or rent your own apartment starting at 800 EUR. As an example, I was first paying 600 EUR (of a 1200 EUR house) and later I was renting a two-bedroom apartment for 750 EUR. You can get an idea of prices and how much space you get in return here.

Other big costs are food (200 EUR), utilities (100 EUR), health insurance (100 EUR), internet/phone (50 EUR), municipality taxes (50 EUR).  You can find much more precise estimates of everything online, such as food. Based on these main expenses, even my past self in 2011, had at least 1400 – (600+200+100+100+50+50) = 300 EUR to save or to spend. Not a “pot of gold”, but definitely enough not to have to budget every expense.

Not part of expenses

Yes, tuition is not on the list of expenses – it does not exist at PhD level. You do follow a couple of courses, but these are paid by the employer.

Also not on the list of expenses are conferences. The general rule of thumb (although this is likely to differ between fields) is that you can go to at least one conference a year, especially if you published a paper there. The registration, travel, hotel, dinner and even the 1-2 glasses of wine you had with dinner are reimbursed.

Paying off student debt is not on my list, either. This is mainly because tuition is low (less than 2K a year) and students (bachelor’s and master’s) used to receive a stipend.  With a part-time job, I didn’t need a loan. This is not the case for everyone, but on average, the debt is 15K, and it’s expected to go up to 21K because the stipend no longer exists. According to the tax office, the average case translates to payments of less than EUR 100 per month.


Next to salary, the labour agreement (friendly English language version) takes care of a number of other benefits that make life easier. First, you have 29 vacation days per year if you work full-time (=38 hours per week). That’s more than 5 weeks of vacation. I’ve never gone on vacation for that long, but I do use vacation days here and there for a day trip, or just to relax after a busy period.

If you are ill or if you are having a baby, you don’t need to use your vacation days – you just get your full salary for up to 39 weeks of illness and 16 weeks maternity leave (fathers only get a few days off, though).

You also automatically build up pension. I have to admit I’ve never really looked into this, because I didn’t feel like there is a reason to worry. While writing this post, I actually looked at my pension account, and discovered that so far, I’ve built up a pension of 220 EUR per month, and I keep working a full-time job, this will grow to 1900 EUR a month when I retire. This is on top of the basic pension (1000 EUR) from the government.

How you see your PhD

Next to the financial side, I feel like the fact that you are employee affects the way you see your PhD. First of all, you are getting paid for becoming an expert at a topic. This is pretty awesome in itself, but it’s also helpful for your self-esteem, even though it doesn’t erase impostor feelings completely.

You and your PI are both employees of the university, with similar employment conditions. Sure, he or she has more responsibilities and more salary, but you have the same rights in terms of leave. You don’t need to negotiate whether you are “allowed” to stay home when you are ill, or if you need to visit a doctor. Of course you should inform the PI, but there cannot be negative consequences of you taking care of yourself.

After 4 years, your contract ends, and you don’t get paid anymore. As a result, people try to finish their PhD on time, and find their next job. Or, since finishing on time is difficult, just find their next job, and plan to finish the PhD later. But the emphasis is that the PhD is a job and it’s normal to move on afterwards.


 All of this helps with what happens outside of your PhD. You just have less things to worry about, so you can concentrate more on the things that are important to you.  You can travel, buy a house, or start a family. Maybe not all at the same time, but the point is, you don’t have to put your life on hold for research.  Hopefully, this translates to a healthier and happier you, and better research as a result.

Firsts: organizing a workshop (part 2)

In the previous post I wrote about getting started with organizing a (sattelite) workshop. In this post I cover a few specific topics that you will want to include in the workshop proposal.

Invited speakers

In most cases a workshop will feature one or more invited speakers. This is something to be arranged early on. Typically already when submitting the workshop proposal, you will need to specify who you plan to invite, and whether the speakers have already confirmed or not.

You might already have a wishlist of people who are famous for their work on the topic of your workshop. If not, it might help to have a brainstorming session, and make an (overcomplete) list of people you could invite. Places to look are:

  • Papers you cite often. Look up the authors, and see if they have recent work, related to your workshop. To do this more efficiently, try SemanticScholar. Here’s what comes up when I search for myself:.Of course, my supervisors are here! The other authors are people in the field whose papers I cite most of the time.
  • Google scholar. Authors on Google Scholar can add keywords under their name. For example, on my profile I have the keyword “machine learning”. By clicking on it, you will see authors who added “machine learning” to their profile, sorted by the number of citations. Note that keywords are author-defined! Therefore, you will not find everyone working on machine learning, and subtopics are not taken into account.
  • Search for your topic of interest and watch some lectures. Bonus: you already have an idea what kind of a speaker somebody is!

To brainstorm, you can create a spreadsheet where all organizers can add potential speakers, roughly with the following fields:

  • Name
  • Website or Google scholar profile
  • Relevance/motivation (i.e. well known for topic X)
  • Whether we have any personal connections
  • Whether the person usually attends the main conference

The last three questions are good to consider, because they influence how likely the person is to respond and accept the invitation. Where the person has to travel from is important because, if you are inviting somebody who wouldn’t normally be at the conference, you probably want to offer to cover the travel costs.

Once all the data is there, you can use the relevance and the chances of the speaker accepting to make a selection of whom to invite first, and who to invite in case the first person declines. It’s also good to decide who will be sending the invitations – usually the organizer who knows the speaker best.


If you want to allow participants to present their work (either as a talk or a poster), there are two main ways to do this:

  • “Type A” contributions, which are novel contributions and which can be published in proceedings.
  • “Type B” contributions, which are abstracts of previously published work, or not fully worked-out ideas and open questions.

Both have advantages and disadvantages. Type B contributions are interesting for people who are already at the conference, but do not have new material to submit to the workshop. Because the threshold for joining is low, you are likely to have more participants. On the other hand, Type B contributions may be problematic for researchers who are not already at the conference (and need a published paper to be able to claim travel expenses). If the conference acceptance rate is low, it’s probably a good idea to have Type A contributions to encourage those authors to participate. A caveat is that you will need to have enough contributions to actually publish proceedings! To combine the advantages of Type A and Type B contributions, several workshops call for both types of contributions.

The mix you choose is likely to influence the schedule of the workshop. If you only have a few Type A contributions, each author could give a talk. If you (also) have Type B contributions, you will probably want to host a poster session. Personally I think poster sessions are great opportunities for the participants to get to know each other, so I would recommend including one in the schedule.

In the workshop proposal, you will likely have to specify what type of contributions you want, how you will collect the contributions (for example, via Easychair) and how you will select the final contributions (i.e. your reviewing process).

Anything else?

By now your workshop program has invited speakers, talks by participants, and a poster session. That’s all, right? Well, that is up to you. Just because most workshops (*at least at the conferences I attended – this might be different in other places!) only feature these building blocks, doesn’t mean that you have to as well. For example, you could also consider:

  • A panel discussion
  • A brainstorming session where participants have to work in groups
  • An ice-breaking activity to encourage discussion throughout the day

Although for the proposal, you probably don’t need to specify a detailed schedule, keep in mind that you also want to leave enough time for breaks, so don’t try to fit too many things in a single day. You might also want to think about organizing lunch (if this isn’t already done by the conference) and/or drinks at the end of the day, so that participants get more chances to interact with each other. This, however, requires a budget – something I will talk about in a later post!

Firsts: organizing a workshop (part 1)

Last week I heard that our workshop, LABELS, was accepted as one of the satellite events at MICCAI 2017. This is not the first time I organize a workshop, but I remember that as a PhD student, I had many questions about workshops. In this post I summarize some things I learnt so far that might help if you want to organize a workshop as well.

What is a satellite workshop?

In my field there are two types of workshops: stand-alone workshops or satellite events.
The stand-alone workshops are more flexible and are basically mini-conferences, but are also more work for the organizers. The workshops I co-organized (FEAST 2014, FEAST 2015 and now LABELS) were all satellite workshops.

Usually, a large conference will have several workshops associated with it. The conference organizers will send out a “call for workshops” or “call for satellite events”, inviting others to submit proposals. The proposal contains information about the workshop topic, the organizers, the invited speakers, how you plan to structure the day (talks, posters) etc.

After the deadline of the call, you wait for a decision. If accepted, the conference typically takes care of the logistics: location, registration, coffee breaks etc. The workshop organizers are responsible for the workshop website, inviting speakers, selecting papers, leading the day itself and publishing proceedings (if applicable).

Who organizes workshops?

Anybody who wants to! Perhaps explaining this is a bit of an overkill, but I do remember thinking that workshop organizers were very well-established scientists, who received special invitations from the conference organizers. When a researcher I was working with suggested we could organize a workshop together, I think I wondered whether we would be “allowed to”.

Of course, the “call for workshops” is already one hint that you don’t need a special invitation. As for being very established scientists, I don’t think that is a requirement. Of course, it’s good to have somebody more senior/experienced on the organizing committee. But it’s not a prerequisite for getting the workshop accepted.

For example, in 2015 as organizers we were one postdoc (me) and two junior faculty, and the workshop was accepted at an important conference in machine learning. In 2016, we submitted a similar proposal (to another important conference), but it was rejected* even though we all had more experience by that point.

*Here I should mention that the acceptance rate for workshops is a bit higher than for papers or grants. You don’t always get to hear the statistics, but in 2014 or 2015 we were among the 90% or so of accepted workshops. The time that our proposal was rejected, the overall success rate was about 50%: much less than 90%, but still not bad.

How to get started?

If you haven’t organized a satellite workshop before but would like to, here are some ideas to get started:

  • Think about which conferences you will probably attend, and study the workshops already organized there to get some ideas.
  • Are any topics missing? Do you know any people who you could team up with, and organize a workshop around that topic?
  • Are there any workshops which have different organizers each year, or perhaps haven’t been organized each year? Contact a past organizer to find out if there are plans for a new edition and offer to help.
  • Ask your supervisor or other researchers (not necessarily at your institution) if they have plans to organize a workshop, and if you can help out.

In the next post I will talk a bit more about some specific tasks related to organizing a workshop, like deciding on the content, inviting speakers, etc. In the meantime, I’d love to hear about your experiences with organizing workshops, and if you have any other tips you can share with others!

Defending propositions: where to find inspiration

Colorful trees in autumn in Japan.

This is the fifth post in the propositions series. If you don’t know what I mean by “defending propositions”, you can read the introduction here.

Now that I’ve discussed a few of my propositions (here, here and here), I thought I’d share a bit more about the process of generating ideas for propositions, and give some advice on how to make this process a bit less painful.

Write ALL your ideas down in ONE place

This might seem like a no-brainer, but it’s very important piece of advice that will make your life easier. I say ALL ideas because the moment you think “could this be something for a proposition?”, you need to capture it. Even if you seconds later realize that probably it’s not, it might give you ideas later in the future. I say ONE place because I kept my ideas in a LateX document, which meant that the writing down only happened when I was using my computer. Of course I made occasional notes on my phone, on a paper I was reading, on beer mats… In many cases I probably forgot that the idea was there. If I had to do it all over again, I would use an app like Evernote.

Find examples

Learning is easier with labelled training data. Find examples of propositions that have been already defended. Unfortunately, like PhD theses, the propositions do not get uploaded to the university library, so they are a bit more difficult to find. A few places to start are:

  • Offices of more senior academics. They accumulate a lot of PhD theses, which, if you are lucky, still contain the loose piece of paper with propositions on them.
  • Blog posts. The best way to start is to look at Project #TweetProp, started by Felienne Hermans. This was continued by a few others, who Eva Langsoght writes about in in this post.
  • Collections of propositions. There are two books (both in Dutch) that I know of: Beste stellingen zijn van hout Paard van Damocles. Although the first book appears to be in stock, I wasn’t able to order it back in 2014, so my guess is that it will be even more difficult now. I did manage to find a copy at the university library though.

Join Twitter

I regret not doing this during my PhD, as Twitter now daily gives me lots of ideas, but it’s never too late to start. You don’t even need to post anything. Just get an account, see who is talking about, for example, #PhDChat, #AcademicSelfcare, #AcademicKindness and follow accounts which shared something you strongly agreed or disagreed with. Soon your timeline will be filled with lots of articles, opinions, memes… you name it!

Think back

Think about advice you received from others, whether it’s in a conversation, email, or maybe a talk you listened to or a book you read (and yes, write it all down). One way to start is to think of a book you liked (it doesn’t need to be advice books, any fiction or non-fiction book will do) and to search for quotes from it on Goodreads. I was really inspired by quotes from Anathem, a fiction novel where one of the major themes is philosophy of science. Since I’m giving advice here anyway, I think reading Anathem might belong to my top 10 “things to do during your PhD” advice.

Get frustrated

Find an (academic) friend, get a coffee or a beer, and talk about all the things that frustrate you. People who only treat you well if they want something from you? Researchers that don’t share their data or code? Reviewers that reject your paper because you reported, you know, ALL the results and not only the best ones? Write them down. Then imagine a utopia in which you can decide how everything in academia gets done. Write that down too.


Don’t think too hard. Let your mind relax, and it will surprise you when you least expect it (or probably, in the shower).

Defending propositions: curiosity and cats

This is the fourth post in the propositions series. If you don’t know what I mean by “defending propositions”, you can read the introduction here.

The proposition

Today I want to talk a bit my last – and my favorite, proposition:

Lack of curiosity killed the cat

The timeline of my propositions shows I only came up with this one at the last moment. After revising the propositions several times, I was a bit stuck, yet I still needed fresh propositions. To get out of my local minimum, I tried changing up my sources of inspiration. Before I relied mainly on my own notes and examples of others’ propositions. My new strategy was to search for articles on topics like “qualities of a good scientist” in hopes this would trigger new ideas. And it did!

I’m sad to say I no longer can find the original article, but one of the qualities it listed was curiosity. I proceeded to have several lightbulb moments as I was attributing behaviors of different people to (lack of) curiosity. But I knew that a statement like “Curiosity is an essential part of doing research” would not cut it, so I had to dig deeper.

I then thought about the saying, “Curiosity killed the cat”, which implies that curiosity is a negative quality. Wikipedia told me that the entire saying is, in fact, “Curiosity killed the cat, but the satisfaction brought it back”. A bit better, but still not very positive. I then did a Google Scholar search on “curiosity” and “cats”, and, lo and behold, found this 1966 (still paywalled!) paper:

In this study, the researchers compared how “curious” different animal groups were. The researchers placed unknown objects in the animals’ habitats and measured how much the animals interacted with the objects, for example by sniffing them. The responses were the highest for primates and carnivores, which included wild cats. The paper then discusses how animals adapt to their environment, for example if their usual food source runs out. Curiosity – considering whether an unknown object could be an alternative food source – could then make the difference between survival or extinction of a species. It is therefore,

Lack of curiosity killed the cat.

My cat Buffy: "Is this food for me?"
My cat Buffy: “Is this food for me?”



To top it off, this proposition helped me have an awesome PhD defense / viva. The first question asked to me by one of my opponents was to explain the proposition. That was great, because I had prepared a slide with figures from “Curiosity in Zoo Animals” in advance. As I started talking, I felt my confidence growing, and it didn’t go away for the rest of the defense, after which I could call myself Dr. 🙂

Defending propositions: an index for reviewers

This is the third post in the propositions series. If you don’t know what I mean by “defending propositions”, you can read the introduction here.

The proposition

I recently wrote about reviewing papers for the first time, and how happy I was when I recently discovered Publons. Publons lets you track your reviewing, essentially making visible all the effort you put into it. Here is an example of a statistic I’ve extracted from my profile:

Example statistic from
Example statistic from

This idea resonated with me because of a proposition I defended a year before, and that I will discuss today.

Introducing an r-index for reviewers as a counterpart of the h-index for authors would lead to more, better and timelier reviews.

Supply vs demand

During my PhD I noticed that the supply vs demand for publications was very different than that for reviews. For publications, supply was clearly higher than the demand: many papers would get rejected, and even if accepted, some papers would never be cited. For reviews, however, the story seemed to be the opposite! It could take months to receive reviews for your paper. And when you would finally get the reviews, regardless of the decision, a few would be very short and uninformative.

I wondered whether this discrepancy could be a result of the reward system. A new publication is rewarded by your name on the paper, by citations, and perhaps even increased chances of getting a position. A new review is – to the outside world – at most a line on your CV. Reviewing is something that is expected to be there, but I doubt it has ever been a deciding factor in a hiring decision. Combine that with well-meant advice from colleagues, “say no to everything, just focus on your publications!”, and you’ve got a supply vs demand problem.


The h-index is meant to measure the productivity of a researcher. Although it has flaws, it is being used — consciously or perhaps subconsciously — to evaluate quality. As a result, it is something researchers can try to optimize for. And for a higher h-index, you need start with getting more publications. Do you already see where the supply is coming from?

What if there was something similar for reviewing papers? If such a – yet to be defined – reviewer index became an important metric, more people would want to review papers and submit their reviews on time, just as they do with papers.

Quality is a bit more difficult. One way to do this is a system where reviews could be seen and discussed by others. This doesn’t mean that the reviewers’ identity needs to be revealed. See for example the reviewing process of ICLR.

Or maybe we can step away from reviews altogether, as Ludmila Kuncheva proposes
here. (The post also features one of my favorite stories about publishing before the internet). Reviews would simply be replaced by citations. Then maybe we wouldn’t need new metrics, and could just go back to looking at the h-index instead.

%d bloggers like this: