The web has much 'data' that is simply not accurate, but you'd think it would be reasonably accurate on matters of simple transcription, where no one has an ax to grind.

Not so, apparently. I was struck by this in Pride and Prejudice,from the Gutenberg project's site: "there are too many in my rank of life who can afford to marry without some attention to money" (emph. added)

The context makes it clear that exactly the opposite is intended: it should be 'not many' rather than 'too many', and this is rather crucial to the tenor of the book. And yes, I found that indeed, many other versions on the web do say "not many".

But here's where it gets interesting. Search for this passage (search for "afford to marry without some attention to money"), and you'll find 32 distinct Google-hits. Of these, 18 have the right version and 14 have the wrong one. In other words, the web has it wrong almost as often as it has it right.

Ugh.

That's amazing wordnerd, though I am not surprised. I have found some terrible medical information on the Web that probably unknowing people access. I especially hate the information that gives people useless ways of treating something serious, rather than telling them to go to the doctor.

Don't blame the medium for the message!

The web has a lot of inaccurate information; it also has a lot of accurate information. The accuracy or inaccuracy is a function of the person providing the data, not the medium that carries it.

The web shares with interpersonal gossip the characteristic of immediacy and lack of censorship. Like the man in the pub who "knows" that NASA never put a man on the moon, the web is a repository for the views of anyone, informed or uninformed, intelligent or stupid, unbiased or prejudiced.

That does not make the web a bad resource; it merely makes it a very accessible one. As with all research it is the responsibility of the researcher to check and double-check all data.

As the curmudgeon next door says, "Some guys would bitch if they was hung with a new rope."

We've spoken elsewhere about mistakes in the on-line versions of the 1913 Webster's Dictionary caused by the OCR software making mistakes that were not picked up before publication on the Web; they were repeated through other sites that used the on-line version as the source for their own dictionaries. It sounds like something similar has happened here. Since project Gutenberg is a well-respected source it it not surprising that the error has been repeated through the Web.

This sort of error is hardly new. In 1631 King Charles I ordered 1,000 Bibles from an English printer named Robert Barker. Only after the Bibles were delivered did anyone notice a serious mistake. In one of the Ten Commandments [Exodus 20:14], a very small word was missed out by the printers. The word "not". This changed the 7th commandment to say Thou shalt commit adultery. That version of the Bible became known as the "Wicked Bible".

The accuracy or inaccuracy is a function of the person providing the data, not the medium that carries it.

Well, of course. Surely no one ever doubted that.

The intelligent, well-educated person will be able to ferret out the good information, versus the bad. Yet, not everyone is intelligent or well-educated. I do think, down the line, there should be some criteria when putting information for the general public on the Web.

I may have spoken too soon in my initial post. This may be a unique, special case.

It seems that about half the google hits, for the erroneous version, come from a single site. And that site has some weird characteristics, ones that suggest that that someone was putting the text on-line, repeatedly, in order to test out some early computer program. Obviously, this would mislead later users of the net -- but obviously, it is not something that would happen often.

Boring details follow, for those who wish them. For examples of the wrong text, google up "too many" "rank of life" "afford to marry" without attention money. You'll get 61 relevant google blurbs (plus three inapplicable ones), a full 30 of which are from "_____.grabafreebie.com". Of those 30, almost every one has a repeated word, saying 'life life' or 'marry marry', etc. And none has a workable link to the actual page with Austen text: its links lead you only to a page saying 'Members Only'. Surprisingly, that happens even if you try the link to Google's cached version,. The cached version will say 'retrieved on Dec 31, 1969 23:59:59 GMT' -- which is rather an odd date.