How To Make It Stop?

Ok, I know that in a post a while ago I said:

I don’t know about you but I melt each of the (very few these days) times I receive a real letter, by post…

So you’d think I’d be delighted with this pile (it is more than three layers deep – and I’ll get at least this many again over the next month or so):


Well, yes and no. What are they?

They are letters written to me applying for postdoctoral positions to work in the group, along with letters of recommendation supporting the applications. This is not unusual. I’m betting that the other guys in my group have their own similar set of letters. Furthermore, anybody in the field who is in even a slightly well known group gets a lot of these at around this time of year. And then there are the email ones as well (ironically, one arrived just as I was typing the last sentence).

Each year, at around this time, a huge number of young people in the field send out such applications. They may apply to anywhere between 15 to 70 or more groups, sending pretty much identical applications to each one, and asking the people who write them letters of recommendation to send a copy of their letter to each of these places.

Now we can have a discussion about why people need to send out so many applications, and what other strategies people use to try to increase their chances of postdoc offers, but that’s not my main thrust right now (but go ahead and talk about that if you want to since there are important issues there). What I am more interested in talking about is why one earth we do things this way at all! Surely, the same field of research that invented the arXiv (for rapid paperless distribution of research papers) can find a way of doing all this that stops so much paper being sent around in multiple copies?

Of the order of a decade ago I suggested (to nobody in particular, just during random lunchtime conversations and the like) that we could fix this with a similar setup to the arXiv, in fact. We have a central database where a person in the field can upload their cover letter, curriculum vitae, research statement, and so forth. The system assigns it a unique identifier. He/she informs her letter writers of that identifier. They upload their letter for that person to the system, and the system links it to that person’s application. Now, all one has to do to get that person’s complete application is go to the database and look up that identifier – either sent to you or your institution by email, or whatever way is preferable. So there is no need to send tons of letters around…. they are only read when needed, etc., etc. Registered research groups are the only places allowed to get direct access to the combined application (materials, cover letter, letters of recommendation). You just read online, or print out and form your shortlist of applicants. In fact, now I think about it, if you’re a registered institution, you can know that someone has applied to you because in their space on the system they can see the list of registered groups and can just click on the ones to which they want to apply. (Sometimes someone will want to tailor their application somewhat differently for particular institutions. No problem… they can have multiple accounts, uploading different versions of their materials if they want.)

Does not seem so hard.

The key point is to make it transparent and universal. More recently, some groups (e.g., European networks in some fields) have started online upload systems (akin to a similar system for graduate school applications that are now increasingly common). This is good, to some degree, but the big problem there is that each of these systems are slightly different from each other. So it has actually made things more time consuming if you are a person’s letter writer. Rather than print off 50 labels and 50 letters all at the same time (and put them into 50 envelopes, and lick 50 envelopes…), you end up havving to go to a host of different websites with different pin numbers and web addresses….. and it seems that it takes longer every year. Makes one long for the days when you just had to worry about where you put that template from last year for printing out 50 Avery labels in LaTeX…

It would be so much simpler with just one database that we all use worldwide, don’t you think? How hard can it be to implement?

Thoughts anyone? What are the flaws that I’ve missed? Most objections I recall from way back -like people being able to easily get access to their own letters of recommendation- are not flaws that don’t already exist in the paper system. (And there are ways around that… and who actually wants to see what people are writing about them anyway?)

letter vs letterBy the way, the other reason I’d like to see the system change? Today, I found in the pile (only been sorting recently now that it is time to look at them) that one of the letters I thought was a postdoc application was in fact one of those rare things: A personal letter from an old friend, sent to me seven weeks ago. You can see from the picture on the right (Click for larger) why it got lost. I’ve put it next to another letter that is in fact a job application. Can you tell which is which?


Bookmark the permalink.

24 Responses to How To Make It Stop?

  1. Supernova says:

    Requesting that application materials be emailed to you is at least a first step toward getting rid of that giant pile of paper. I’m in the middle of the whole daunting application process right now, and I’m mystified by the applications that still want hard copies of everything (sometimes multiple hard copies of everything). Is it so important to have a physical signature on a recommendation letter?

    I do like your idea of a central job site. I think an obstacle to its implementation is the very decentralized nature of the field. Who would set up, run, and maintain such a site, and what would be their motivation for doing so? It seems vastly more complicated than simply maintaining a list of job ads, as many of the professional societies do.

  2. Aaron Bergman says:

    Having just done a little job hunting myself, I’ve discovered that that’s pretty much what the math people already do. It’s called Pretty cool if I do say so myself.

  3. Alejandro Rivero says:

    One first step could to be to stop asking recommendation letters to be sent in advance. First do the shortlist, then ask the shortlisted to send recommendation letters.

  4. Clifford says:

    Wouldn’t really be a shortlist. I for one need letters to help me decide who should be on any shortlist. The typical candidates’s CVs alone at that stage in their careers are not always enough to help distinguish them. So you need the letters at the very earliest stages.

    Also, email applications are not an improvement, really, Supernova. And as to your last paragraph…. the arXiv is an example o fsomething that exists and works really well despite all those questions being relevant to it at its inception also.


  5. javier says:

    For me, as a prospective postodc, such a system would be great…
    Still, some places require me to send not only cover and recomendation letters, but also hard copies of all my papers by regular mail. Even if they are avaliable at the arXiv.
    One problem with your arXiv-like proposal is that you are making certain information avaliable to everybody. I don’t know how you do it in physics, but in math usually the referee send the recomendation letter directly to the institution, not to the applicant. I don’t know how many people would write a honest recommendation letter knowing that anybody can read it.
    There is one somehow similar idea at, where any researcher can display his cv, and any institution may announce any open positions. It doesn’t have much of a hype, though…

  6. Warren says:

    The difficulty isn’t in the software, it’s in the people.

  7. Bee says:

    Hi Clifford,

    that would certainly be a good idea, and also make life much easier for many postdocs!
    CERN as well as PI do their applications online, which is a nice thing once you get used to it (I didn’t initially like it), and it works faster.

    Just some remarks:

    1) there are definitely cases when you want the cover letter to be somewhat personal, like e.g. someone you’ve met at a conference and you want to refer to this (make sure he remembers you).

    2) One thing to take care of: confirm, confirm, confirm, and let the people know if they didn’t get a confirmation, something went wrong. It’s so easy with online forms to completely mess up things. E.g. PI this year hasn’t only changed the websites but apparently also the application listing (not that I could tell whether it has changed), but on several occasion people apparently loaded up their files, didn’t get any error messages, but the files never showed up in the listing. This is a potential disaster! (same with emails, there’s always a chance the email got lost, spam-filtered, bounced back to nirwana or whatever, but you don’t find out until some weeks later someone is confused when you ask why he did never reply)

    3) If sounds to me similar to replacing job-offers with searching for job ad’s. If I recall that correctly, tip-top has a section where you can advertise yourself with a title like: LOOKING FOR POSTDOC POSITION, ANYWHERE, ANY SALARY, WILL DO EVERYTHING YOU WANT! I don’t know of any occasion somebody was contacted as a result of such a posting. If it was more common that people offering jobs would read ads from people searching jobs this could decrease the amount of letters you get send.

    4) You need some human being who supervises the system and acts as an emergency contact. E.g. I’ve once successfully filled out an application and managed to mistype my email twice with the same mistake (ubsc instead of ucsb). I got send a password to access my files. Which I never received, obviously. I couldn’t just fill out another application because the system found somebody with my name had already applied… what then?

    5) Alejandro: nothing goes without letters. How are you supposed to judge on 100 people you’ve never meet, who’ve written only 3-5 papers, have a CV which fits into 2 pages, and a research proposal like everybody else’s? (I didn’t fully realize its so hard until this year)

    6) For those who are currently out there looking for jobs, see also: A Hazy Shade of Winter



  8. Here’s an optimal situation that I imagine for the details. It wouldn’t be hard to design this system so that you can personalize your application for each school. It would be sensible if the system was set up so that you can upload many PDF files to your account: different personal statements, different research proposals, different cover letters, etc.

    The process of applying to a school would then simply be a matter of choosing the school and the position available at the school from a list, and then selecting which documents in your account will be made available for that application.

    Users of will recognize the inspiration for the following. To get your recommendation letters into the system, your writer would need to have an account on the system. You would then look him up in the user directory and click the option to send him a “Letter Request.” He will get an email with the subject line “John Doe has requested a letter.” When he logs in next, he’ll click the “Upload a letter for John Doe” link on his user homepage, which will then allow him to upload his PDF reference letter. It will ask him if he wants to give Mr. Doe permission to read it.

    An applicant could have letters from more than the standard three recommenders. If he so desired, it would be very easy for him to use different combinations of recommenders for different schools. For him, it’s just a matter of selecting which documents are being sent for a particular position. For the recommender, it’s all the same (after uploading the letter, everything is out of his hands).

    The system could be used for more than just post-doc/professor positions — also for summer institutes and such things that require letters. And –why not– even for grad school applications. Perhaps people reading the applications would see the major disadvantage being that they will receive many more applications, since it’s so much easier to apply.

    The degree to which this system would ease the lives of so many people is beyond measure. But instead of simply talking about it, let’s move this discussion towards actually getting it developed. I’m just a student with no money or web design experience, but I’m sure there are a number of big shot professors at big shot schools who read this blog and can take some initial steps.

    Clifford, maybe you or someone else can get funding from your insitution or elsewhere to begin such a project? I don’t know how that sort of thing works, I’m just throwing out ideas.

  9. Alejandro Rivero says:

    I known a guy in humanities that was recruited into the CSIC because the team, it seems, was used to ask about the young people in every campus they were giving a talk. And how many universities does a research team visit along a year? The result was that they were able to fish out high quality students even before the student were asking for a grant. I am not telling that this is the method to follow, but it is a counterexample of the proposition “everything must be done as it is done today”. I am sure the letters are needed to take the final decision, but are they checked even before checking the CV and the research plan? Are the letters checked or just the name or institution? Because if the latter, the candidate can be asked during the shortlist phase to produce proof of the existence of such contact, and he can even be blacklisted if he is found to falsify his contact list, or penalised if he is not able to produce the letter.

  10. Peter Woit says:

    Kishan and others,

    I think it’s worth repeating Aaron’s comment. The math community has already done this, creating a system that does exactly what you suggest (see This was originally a project of the Duke University math department, and the software was written by one of their people, Yunliang Yu. Responsibility for it has been taken over by the AMS, and they have started charging a $350/year fee to institutions for using it. That price covers ads for up to seven different jobs from a department. If a particle theory group wanted to use this, I suppose they could right now (already some physicists are using the system to apply for jobs in math departments), but it would make more sense for the APS or some other group to acquire the software from the AMS, customize it and set up their own system.

    Columbia has been using the system for several years now and it works well. One of its virtues is that it allows any department member involved in hiring to have access to most of the application materials from any computer anywhere. No more running around trying to find someone’s application folder. You can sort on any number of fields, so can quickly pull up applications by sub-field, by date of Ph.D., by Ph.D. institution, etc.

  11. yagwara says:

    Seconding Peter’s comment. As a mathematician on the market this year, I can tell you that is a joy to use. It puts everything in one place, it is intuitive, and I have never found it to be buggy or erratic.

    You physicists got us hooked on arXiv, now we can finally return the favour!

    My one complaint: what is with the ~50% of math departments that don’t use

  12. Clifford says:

    One comment I’d like to make about things so far: I don’t think that the system should be entirely about jobs that have been advertised. Sometimes you want the unsolicited applications that flood in. This is because sometimes you don’t know whether you’re going to have a job available from one day to the next. I like the system now that you have a hundred applications on file, and you hear of an unexpected source of money and then you can go to your box and find a candidate. To build the system such that it is based only on people sending applications for an advertsied job would remove this flexibility. I think that people should just be able to go in and click the institutions/groups they’d like to send an application to and leave it at that. Then you can just go into the system and look and see who has added you to their list of places they would like to work.

    And yes, it does sound like the mathematics community’s system might be a good start. I’m not comfortable with the fee system though. I think it would be nice if it were a simple service to the community along the lines I mentioned in the post, as the arXiv is.

    Someone mentioned something about the applicant being able to see their letters. I don’t see why they get to see their letters. Only the people reviewing applications to their group will be able to see all of the application materials.

    Bee, of course a human being will be in the background for emergencies. Same as when something goes wrong with a submission to the arXiv.


  13. Alejandro Rivero says:

    My one complaint: what is with the ~50% of math departments that don’t use”

    Well it is obvious that can offer a service to print and send via ordinary mail the letters and CV in the database, paid either by the institution or, at low cost, by the user. Actually no mail is needed, they could email the package as pdf to the “foreign address”.

  14. Moshe says:

    I agree with everything said so far, it would be wonderful to get rid of the piles of papers, the tedious job of sorting them, copying them etc. etc. Also, like for faculty jobs, multiple electronic systems will only make things worse…

    Let me add one more detail: as a newish string group we have struggled getting ourselves into the collective consciousness. How do you advertise your job when nobody really reads any advertisement and everyone relies on a few mailing lists floating around? to get yourself on one of those lists takes some concentrated effort, I think by now we broke the code…seems to me like the most conservative approach possible (in my definition of conservative, which is resistant to change…).

    (in other news, Firefox 2 has built in spell-check, goodbye to all the exotic animals and such…)

  15. Taking Peter’s suggestion, how do we now get someone at APS to map this from talk into reality? (Again, I’m looking for hot shot physicists to come forward…)

  16. Clifford says:

    Kishan Yerubandi: – I doubt any hot shot physicists trouble themselves to read my blog. We can hope that maybe some cold or warm shots might tell some of the shots at successively higher temperatures… until it gets to the ears of someone looking for a project, or who can allocate the resources.

    I just hope that it is done in a way that mimics the arXiv more, is not tied to specific job ads (see my previous comment for why), and leaves money out of it as much as possible. That way it is a level playing field for all groups worldwide. Like the arXiv did for publishing ideas.


  17. FineStructure says:

    I was wondering if you receive similar volumes of letters from prospective graduate students?

  18. Clifford says:

    Some volume, but there is a well-defined system and infrastructure to handle most of that. It works reasonably well. The postdoc system in our field, producing the stuff I gave a picture of above… that’s clearly broken.


  19. Chad Orzel says:

    Here’s a Manual TrackBack to my comments on this.

  20. alienmist says:

    Like one very religous man said… put yourself in the shoes of your neighbour( ..Well the applicant’s shoes).

    For some strange reasons once people make it, they forget all they did to get where are at the moment!

  21. Supernova says:

    email applications are not an improvement, really, Supernova.

    Why not? Your goal is to “find a way of doing all this that stops so much paper being sent around in multiple copies”. Multiple virtual copies are a big improvement over multiple paper copies, in my opinion. I don’t see much difference (at least in quantity of paper) between references sending emailed letters to X number of hiring committees and X number of hiring committees downloading virtual letters from a website.

    An intermediate step might be for each group or institution to have its own site where prospective applicants and letter-writers can upload documents (whether in response to advertised positions or just for general consideration). Sure, then each applicant/letter-writer has to do it more than once — but I’ve been tailoring everything to the individual institution anyway, so for me that wouldn’t represent any more work. Some fellowship competitions in my field are already doing this, and it seems to work well, though one annoying thing about it is dealing with the cutoff time for completed applications.

    Some of the postdoc and faculty applications I’ve done this year have even gone through their universities’ HR department online application systems — these are generally reasonably easy to use, but kind of impersonal. Also, they tend to have lots of bureaucratic components where you have to check boxes certifying (for example) that you have a Ph.D. or that you’ve never worked at that university before. Seems like a department in a technically-oriented field could easily set up its own system and worry about the HR stuff once they’ve made an offer.

    I do like the arxiv model in principle; just thinking about other (possibly easier) ways to at least partially address the problem.

  22. Clifford says:

    It is not just about quantities of paper. It is about centralizing the information where it can be used most readily. Having many components of a single application (from as many as four or five people – the applicant, their letetr-writers, etc) existing in multiple formats on some member of a group’s email inbox is not a good solution. This is why databases were invented. We should use them.


  23. Mark Stern says:

    This year, in response to the overwhelming success of, the Duke Mathematics Department launched a broader service, This is designed to provide to other disciplines essentially the same service that provides to mathematics.
    You can learn about the service at .

  24. Randal says:

    As much as this seems like madness, people are only responding to the forces of competition as they understand it. While it might be easier for all concerned to submit their applications using some electronic mechanism, anyone who sends their by regular post will have an advantage. As you’ve pointed out, physical mail has appeal in a world inundated with email and spam. I think you can expect candidates to use paper for as long as they think it will give them an edge.

    Similar dynamic: I go to conferences all the time where the event planners have declared the dress code to be business casual. But everyone knows the few that show up in ties will stand out among their peers. So everyone shows up in a tie, and no one has an advantage.

    The only way to make the electronic alternative work would be to make electronic submission mandatory for everyone. I don’t think we’re there yet.