I think there are two interesting questions here: 1. What is a number? A magnitude, an ordinal value (obeying the transitive property), a rotational value (like day of year, degrees, day of week), an interval value, a nominal labeling (take the SS Stevens taxonomy and add rotational [1])? Addresses frequently, but moreso in cities than in rural areas [2] have the property that 123 Huaihai Zhong Road is geographically between 120 Huaihai Zhong Road and 130 Huaihai Zhong Road, hence obeying the transitive property when articulated into geography. 130 State Street SW < 30 State Street SW < 30 State Street NW < 130 State Street NW, meaning that sign is written using notation other than the minus sign. Generally, in places with European cultural heritage, the divisibility of the number (by two) determines the side of street, though even in Europe there are some fun and remarkable differences [3]. Many times, as with the ordinal numbers (first, second and so forth) there is the assumption that the n-th president will have served during a time intermediate between the n-1st and n+1st president. That is, inferences, in the classical sense of the semantic web, may be drawn about many classes of entities considered to be numeric. Do all things for which a simple bijection between the elements of a set and the integers, inherit the "number" property of that bijection, or simply if the bijection is humanly intuitive (though the cardinality of the rationals is aleph null, we might not expect the standard Cantor labeling to convey the ordinality to such). It is reminiscent of work with the gravitational flavorings of graphs [4] in which we ponder the question of how simple graphs like the internet might be flavored with a single dimension of artificial gravity so as to guide simplified navigation.

2. What sort of interface is best used to elicit a numerical response from a person? We often assume that the human will type such a thing, though for small n, radios and even selects work okay. Can a widget be developed called a "throttle" which allows the user to use a joystick to accelerate through an ordinal collection of n ordinal values (for large n>10000) and to pick a number more quickly using the throttle than by using the keyboard? Since the words of an alphabetic language have a natural ordering (imposed by alphabetization) are not words numeric, and cannot a throttle be used more effectively than a keyboard? Cheers David [1] http://en.wikipedia.org/wiki/Level_of_measurement [2] http://bitboost.com/ref/international-address-formats/prc-china/ [3] http://en.wikipedia.org/wiki/House_numbering [4] http://www.graphicalweb.org/2012/#presentation_19 -----Original Message----- From: whatwg-boun...@lists.whatwg.org [mailto:whatwg-boun...@lists.whatwg.org] On Behalf Of Nils Dagsson Moskopp Sent: Saturday, March 08, 2014 1:25 AM To: Jonathan Watt; whatwg Subject: Re: [whatwg] <input type=number> for year input Jonathan Watt <jw...@jwatt.org> writes: > is it wrong to use <input type=number> for year input. I am certainly not an expert on the topic, but I believe the conceptual problem can be reduced to using an input designed for a group (in the mathematical sensce) to represent a value that is torsor. Quote <http://ro-che.info/articles/2013-01-08-torsors.html>: > While adding two dates is not possible, it is possible to add a time > interval to a date («five days from today»). This suggests that we > should not confound dates and time intervals — they are different > types of values. Therefore asking for a duration using <input type=number> is fine – asking for a calendar year, however, is obviously a type error. <http://math.ucr.edu/home/baez/torsors.html> -- Nils Dagsson Moskopp // erlehmann <http://dieweltistgarnichtso.net>