Alphabetical order has nothing (or little) to do with meaning. Classifying and arranging the OED's word-stock according to its respective definitions, on the either hand, creates what the HTOED editors call 'conceptual maps', which can then be further analysed to tell us both about language and about the referents of language. It is self-evident that the potential uses for these 'maps' are far-reaching, and that the creation of HTOED is of great lexical and cultural significance.

HTOED re-organizes OED's vocabulary under three main headings: the external world, the mind, and society, with every word and sense allotted its slot in a nested series of sub-categories which extend many layers downwards. In this way, the thesaurus allows us to see, grouped together, words relating to a comprehensive range of cultural phenomena, with dates of first usage and evidence of currency over specific periods. See further (for reviews) Brewer 2010 and Mugglestone 2010.

The printed form of the HTOED is unwieldy and prohibitively difficult to use: one has to turn back and forth between two heavy volumes, scanning closely written lists of many-digit numbers, in order to identify what one is looking for - i.e. the collection of words and phrases semantically close to the word one is looking up. Electronic linking between the OED and the HTOED is therefore a wonderful boon: a few clicks can now achieve what in the past required much tedious and time-consuming labour.

HTOED and OED Online 
On OED Online one can access HTOED either direct or via individual dictionary entries, most of which now contain one or more links to the corresponding entry in the thesaurus. Clicking on a link marked 'Thesaurus' within an entry causes a small box to spring into view, with a list of related terms (themselves clickable) ordered by date of first use. Thus if one looks up the verb rub (Third Edition March 2011), for which the lexicographers identify 11 separate main senses, several additional senses, and many different phrasal senses, one finds that almost all of these have an individual link to HTOED. So the link for sense 1 ('To apply pressure and friction to', etc) gives access to the following list of verbs, under a heading which locates the sense within HTOED's taxonomic structure:

Related words: 

• gnide (c1000)
• frot (?1225)
• gnodde (c1230)
• rud (c1300)
• rub (a1325)
• wipe (1362)
• fret (a1400)
• wrive (1481)

In turn, one can look up these other words and discover their respective definitions and dates of use. Clearly, such information has many different possible uses and applications; in particular, it is tempting to make inferences on the currency of words or senses at different periods, e.g. the number of available synonyms at any one time.

Problems
But there are significant limitations or problems with the HTOED linking which users should be aware of. The thesaurus as originally compiled was based almost entirely on OED2, so it reproduces many of the biases and imperfections of that edition. Most obviously, dates of first usage are insecure for unrevised OED entries: so in the above list, frot, gnodde, wipe, fret, and wrive are all as yet unrevised, and MED evidence antedates wipe, for example, to a1225. (Happily, OED Online automatically updates the thesaurus dates for words which OED3 has revised).

There are more serious problems too. Because over half of the OED is as yet unrevised from the first edition of 1884-1928 (the latest figure, as of December 2011, is 37% revised, i.e. 63% unrevised), it is not surprising that the HTOED link delivers uneven and incomplete results on many types of vocabulary. This can be seen by looking at terms where OED1 (and hence OED2) coverage was notably deficient, and where OED3 has been making substantial changes; for a short study of terms for lesbian go here. No guidance is offered on the OED Online website to alert users to possible pitfalls; this is an urgent desideratum.