Multi-district preference modelling

This paper with longtime coauthor Geoffrey Pritchard is an important step toward systematic design of electoral systems.

Abstract: Generating realistic artificial preference distributions is an important part of
any simulation analysis of electoral systems. While this has been discussed in some de-
tail in the context of a single electoral district, many electoral systems of interest are based
on districts. Neither treating preferences between districts as independent nor ignoring
the district structure yields satisfactory results. We present a model based on an extension
of the classic Eggenberger-Pólya urn, in which each district is represented by an urn and
there is correlation between urns. We show in detail that this procedure has a small num-
ber of tunable parameters, is computationally efficient, and produces “realistic-looking”
distributions. We intend to use it in further studies of electoral systems.

New algorithms for matching problems

A preprint with Jacky Lo, just submitted to arXiv and on my publications page. The paper was ultimately inspired by a Christmas party game, which shows that the line between work and play is hard to draw.

Abstract: The standard two-sided and one-sided matching problems, and the closely related school choice problem, have been widely studied from an axiomatic viewpoint. A small number of algorithms dominate the literature. For two-sided matching, the Gale-Shapley algorithm; for one-sided matching, (random) Serial Dictatorship and Probabilistic Serial rule; for school choice, Gale-Shapley and the Boston mechanisms. The main reason for the dominance of these algorithms is their good (worst-case) axiomatic behaviour with respect to notions of efficiency and strategyproofness. However if we shift the focus to fairness, social welfare, tradeoffs between incompatible axioms, and average-case analysis, it is far from clear that these algorithms are optimal. We investigate new algorithms several of which have not appeared (to our knowledge) in the literature before. We give a unified presentation in which algorithms for 2-sided matching yield 1-sided matching algorithms in a systematic way. In addition to axiomatic properties, we investigate agent welfare using both theoretical and computational approaches. We find that some of the new algorithms are worthy of consideration for certain applications. In particular, when considering welfare under truthful preferences, some of the new algorithms outperform the classic ones.

Fair Open Access Principles

I have been working for the last 18 months with a group of talented and committed people to accelerate conversion of subscription journals to open access. There are many barriers, and many pitfalls. For example, so-called “predatory” open access journals that take authors’ money and provide no quality control have gained considerable publicity and must be avoided. Large, inefficent and greedy commercial publishers have attempted to “double-dip” by introducing Hybrid OA. Otherwise well-run open access journals still have high publication charges.

Our aim is to avoid these problems by retaining community control of journals and adhering to high ethical standards. Here is a list of our basic principles, based on the original version introduced by LingOA. This list was developed after extensive discussion and some consultation with other OA advocates such as Peter Suber and Marie Farge. We hope it will be useful in delineating what we see as the ideal way to publish journals. This is not to say that all other ways are necessarily “unfair”, of course, although some of them clearly are!

The Fair Open Access Principles

  1. The journal has a transparent ownership structure, and is controlled by and responsive to the scholarly community.
  2. Authors of articles in the journal retain copyright.
  3. All articles are published open access and an explicit open access licence is used.
  4. Submission and publication is not conditional in any way on the payment of a fee from the author or its employing institution, or on membership of an institution or society.
  5. Any fees paid on behalf of the journal to publishers are low, transparent, and in proportion to the work carried out.

Clarification notes:

  1. This could be ownership by an editorial board, or by a democratically controlled scholarly society, for example. Key points are that the controlling organization, not a commercial publisher, must own the journal title, so that a change of service provider can be achieved without changing the title, and so publishing companies simply compete to offer services to the journal. We strongly recommend that the ownership structure allow for democratic input by the community of readers, authors and referees, in addition to editors, and that procedures for making key decisions about the journal’s future be formally (even legally) specified. We strongly recommend that the governing organization be fully nonprofit (for example, IRS 501 (c) (3) in USA). A for-profit company accountable only to shareholders is not compatible with Principle 1.
  2. The  journals and their publishing house can still propose, among their services, to take care of possible legal issues pertaining to copyright on the author’s behalf, under the author’s oversight. We strongly recommend that reviewers also retain copyright of their reviews, and journals retain ownership of all correspondence and mailing lists compiled on the electronic submission system put at their disposal by the publisher.
  3. Any form of subscription paywall is unacceptable, including “hybrid OA”. We  strongly recommend that the industry standard CC-By licence be used. All content of the journal should be easily accessible from the journal website to anyone with a standard internet connection.
  4. The key idea is that journals be “free at the point of use” by authors and readers. Principle 3 deals with readers and Principle 4 with authors. Compulsory APCs (article publication charges) are not compatible with this principle. Journals should ideally be funded by general contributions from universities and research funders, with these contributions not tied to individual articles or groups of authors. Principle 4 is not compatible with “APC Big Deals”, whereby institutions pay for APCs of their employees but do not contribute to a general fund. Also not compatible is the practice of charging APCs by default to the author’s institution, with waivers for authors who do not have institutional funds. The principle does not preclude voluntary APCs, but requests for these must be unobtrusive and no barrier to publication. APCs must be “opt-in”, never “opt-out”.
  5. “Low” depends on the particulars of each journal, but we strongly recommend an absolute maximum of $1000 per article published or $50 per page for the total expense of any journal, and substantially lower fees in all possible cases. We recommend that an itemized price structure be made public in order to ensure transparency and make the proportionality principle apparent.


A rant about Lufthansa

Apologies, a rant is required. I just swore at someone on the phone, which is quite a rare occurrence. Read on to see why, if you dare.

Lufthansa is by far the worst airline I have ever had the misfortune to deal with (I haven’t flown on anything really unsafe, of course, but have flown in Iran on a Tupolev). Not only is their in-flight service embarrassingly behind my usual carriers such as Emirates, Singapore, Air New Zealand, their customer service is a joke. My wife has flown them twice in the last year.

Example 1: The first time she bought an in-class upgrade in order to make a further upgrade using Mileage Plus miles. Then was told that the upgrade was not possible. Nor was a refund and in-class downgrade. She flew in the same seat as before, and after 10 months they have offered to refund 20% of the cost, but still haven’t done it more than 3 weeks after asking for bank details. Each time I enquired it took weeks or months to get an email reply.

Example 2: She was denied boarding for LAX-Munich on a trip from Auckland to Europe (because of a newly enforced rule on passport validity that was not picked up at Auckland), and the other 5 legs of the trip were then cancelled. She had to buy a 1-way ticket home. Note that because my wife never made it to her invited conference talk, she doesn’t want to ask the organizers for any money, and they probably think she is being refunded by the airline. The original was bought through Expedia. After hours talking to Expedia we have been told that Lufthansa will within a few days let us know how much they will refund, and then it will take 10 weeks for a refund. I asked Expedia to transfer me to Lufthansa in order to discuss the extra costs incurred that were not on the ticket. The LH staff member told me this number is not for customers, only travel agents. She refused to give me any contact numbers other than the general US 1-800 number. Not wanting to start all over again and failing to convince her with reasoned argument I yelled and hung up.

I don’t want to make too many jokes about Germans, but there is some really stereotypical behaviour going on here. Is customer service, or basic human flexibility, such as difficult concept to grasp?

I strongly urge anyone reading this to avoid Lufthansa if possible. Other tips – never buy your own ticket if it is supposed to be reimbursed by others, get them to buy it directly; make sure your passport has 3 months validity past the end of your stay in Schengen area. And maybe consider cancellation insurance!

What happens to journals that break away?

Although it is still a relatively rare occurrence, several journal boards have broken away from large commercial publishers. A good list is at the Open Access Directory. These journals usually are required to change their name, because the previous publisher will not relinquish it. They are cut off from the enormous support provided by large commercial publishers (after all their subscription prices are so high, the money is surely being put back into developing better infrastructure, rather than, say enriching shareholders, giving inflated honoraria to editors or paying inefficient support staff). Thus one might expect that these journals would struggle.

I looked at the fortunes of the mathematics journals that have taken this route. Below I list the original title name, the approximate date of the breakaway, the new title and publisher, and citation impact measures taken from 2014 data at, and compare them to the results for the original journal. Those measures are EF (size-dependent measure of importance) and AI (analogous to Impact Factor, but based on the same kind of reasoning as underlies PageRank – not all citations are equal). Each has a maximum value of 100. These are of course not the only measures one could use. I also list CE, the 2013 cost-effectiveness rating from (essentially, subscription cost per citation) – the smaller the better.

Old: Journal of Logic Programming (Elsevier), changed name more than once to Journal of Logical and Algebraic Methods in Programming, still publishing, EF = 0, AI = 0, CE = 84.73
New: (1999) Theory and Practice of Logic Programming (Cambridge), EF = 31, AI = 40, 42.33

Old: Machine Learning (Springer), EF = 77, AI = 92, CE = 27.01
New: (2001) Journal of Machine Learning Research (diamond OA), EF = 94, AI = 97, CE = 0.0

Old: Topology and Its Applications (Elsevier), still publishing, EF = 78, AI = 33, CE = 32.34
New: (2001) Algebraic and Geometric Topology (Math Sciences Press), EF = 77, AI = 77, CE = 3.67

EDIT: I received email from Alex Scorpan saying: “The facts are that AGT was born by splitting off from “Geometry & Topology”. The resignation of the board of “Topology and its Applications” may have occurred at the same time, may have involved people on the board of AGT, and may have involved the same ethos that moved the founders of GT and AGT, but otherwise the two events are not connected.” Alex has edited the OAD wiki to fix this. I haven’t looked into the question any further.

Old: Journal of Algorithms (Elsevier), stopped publishing after 6-7 years.
New: (2003) Transactions on Algorithms (ACM), EF = 60, AI = 76, CE = 5.57

Old: Topology (Elsevier), stopped publishing after 6 years
New: (2006) Journal of Topology (Oxford), EF = 70, AI = 93, CE = 39.24

Old: K-Theory (Springer), stopped publishing very soon, and archives disappeared.
New: (2007) Annals of K-Theory (Math Sciences Press) (after an intermediate change to Journal of K-Theory (Cambridge), EF = 59, AI = 79, CE = 102.47), too new for EF, AI

Old: Journal of Philosophical Logic (Springer), still publishing, no EF, AI or CE listed (the website lists only “interim editors”)
New: (2007) Review of Symbolic Logic (Cambridge) EF = 35, AI = 49, CE = 222.58

It seems clear that the new journals are doing considerably better than the old ones overall. I wonder whether the idea often touted by radical leftist OA advocates that large commercial publishers don’t add much value could have a grain of truth in it.

A blockage cleared

After a very long time, some (revised versions of) papers have emerged, two on distance rationalizability of voting rules with Benjamin Hadjibeyli, one on signed networks with Samin Aref, and one on “wisdom of crowds” with Patrick Girard and Valery Pavlov.

There is a long list of things still to do, but this is a big relief. These papers had gone through substantial revisions and years of work in some cases. There is also a recent paper on legislation networks with Neda Sakhaee and Golbon Zakeri.

My papers can always be found on my publications page.


I saw a concert version of this tonight with Simon O’Neill in the title role. While the plot is quite weird and distasteful in places, the orchestra and singing were excellent. I saw Simon O’Neill in 1997 (I think) in a small role in the play Master Class in Auckland. He has certainly gone up in the world!


I attended this biennial meeting for the fourth consecutive time. Attendance in Toulouse was substantially larger than previously. Organization was excellently led by Umberto Grandi.

Toulouse was busy with tourists augmented by fans of Euro 2016 football teams and evenings were very noisy downtown. The city seems like a pleasant, slightly provincial place.

There were many interesting talks but I find it hard to maintain concentration. The poster sessions were surprisingly interesting and perhaps in future there should be much more time spent on these, and shorter talks given.