Bad is probably too strong a word, but MARC has severe failings, most of which seem actually to date from its conception, rather than being exposed by changing practice. That said, the main failing seems to be that it was not designed to be used with OPACs and this seems to have shaped its shortsightedness. I have used the shorthand $ to refer to subfields (e.g. $a to refer to subfield a). All references are in fact to MARC21 (formerly USMARC) which is now the de facto standard since the demise of UKMARC, which was arguably better in many areas, and only MARC21 Bibliographic (rather than Holdings, Authorities, etc.). Information on specific MARC fields can be found on the online Concise manual. I’m sure there are more things: please add to the list.
- It uses totally different organisation to AACR2, which it MARC an attempt to encode. AACR2 is nothing if not scrupulously numbered and organised, subdivided into tiny rules. The best examples for this can be found in the notes fields (5XX). Notes in AACR2 follow a prescribed order and are logically numbered; in MARC, notes are numbered, but in a totally different order and often not at all, leaving far too many notes in the general note field (500). When cataloguing videos, for instance, AACR2 has the cast note first then the crew note; MARC made the cast note 511 and the crew note 508. Persuading cataloguers to put the notes in reverse numerical order, as should happen, is not easy.
- MARC encourages cataloguers to create a good display, rather than think logically. AACR2 runs through the sources of information, the description of the item, then the application of headings. By making the 100 field come before the 245, MARC encourages cataloguers to make the main entry decision first.
- MARC relies on restricted numerical sequences to encode information, making much the same mistake as Dewey when he confined himself to 10 digits and put the notation before the logic of the scheme. A good example is the 246 which uses the first indicator to show the kind of added title concerned. This means that only ten possibilities are allowed. Most will end up with first indicator 3 for Other title. There are only a certain number of avaible fields, i.e. below a 1000: this should be enough, but the 500s are getting rather choked, and may be a preventative to any scheme of putting them in the right order. Another example are the indicators, which are undoubtedly useful, but are restricted to two in number, so that when mnemonic meanings clash, confusion ensues: the 2nd indicator of the 245 and 440 fields does the same job as the first indicator of the 740. The confusing mess of the fixed-length fields (LDR, 008, etc) does not help. Numbers do make it more international (except if you don’t use Roman numerals of course) but the AA of AACR” stands for Anglo-American and the creators of various MARCs have till now been too concerned with internationalism.
- It duplicates and sometimes even loses the effect of ISBD punctuation as prescribed in AACR2. For instance, the place of publication, publisher, and date of publication have clearly defined punctuation: Place : Publisher, Date. These three elements are put in subfields $a, $b, and $c and the punctuation must be put in after each element: $aPlace : $bPublisher, $cDate.. Why have data elements defined twice?
- MARC sometimes even loses the effect of ISBD punctuation. Take 245 $b which can mean Other Title Information or Parallel Title: whereas the ISBD punctuation makes this clear with either a comma or an equals sign, MARC uses $b for both. With complicated records, entire parallel titles, subtitles, and statements of responsibility can end up bundled up in a $c, which is meant to hold the statement of responsibility but which luckily has an “etc” in the usage notes.
- MARC allows you to do things which are technically not allowed. A good example is, ironically, uniform titles. AACR2 is clear on the point that the initial article should be omitted from uniform titles (rule 25.2C). This means The Bible becomes Bible. However, the the second indicator of the 240 field allows you to specify the number of non-filing characters. Why? The whole point of uniform titles is that they are uniform. I can only think that this was allowed to accommodate the policies of a particular library, but if it is contrary to AARC2, and especially when more and more catalogues are being combined in various ways (e.g., OCLC Worldcat, the RLG union catalogue, and COPAC, it seems unwise to differ unless you really have to. My personal beef is the 533 field which allows you to create a catalogue record based on the original of something like a photocopy, then add a note saying (something like) “this is a photocopy” even though AACR2 instructs you to catalogue the copy and provide a note on the original. This has particular ramifications for things like ebooks, where most records are being created the 533 way, often by computer, and mainly because netLibrary, part of OCLC and therefore close to the Library of Congress, persuaded LC to introduce a Library of Congress Rule Interpretation to let them do it. A rant on LCRIs and the maintenance of library standards is probably better left for another time.
It seems that MARC is unsure of how detailed it wants to be and how it wants to store information. There is a fair case, I think, for lumping all the 245, 260, and 300 in one field as description which can be searched by keyword. The headings should provide the entry points (including a specific field purely for adding the main title. This might present problems with brief display and such like, though it might be possible to extract information from the super245 by using the metadata delimiters- the punctuation- already included in an AACR2 record. Failing this extreme solution, AACR should be redesigned at the same time and in the same process as MARC, so that they are completely integrated, so that tags and punctuation are not doing the same job, so that OPACs are free of the tyranny of card catalogue records, and so that cataloguers can concentrate on making the correct decisions for description rather than playing around with irrelavent numbers.