Wreck havoc

[Chris Waigl]

Damien Hall wrote:

>Spotted in *Metro* (Philadelphia edition), 15 September 2005, p7 (letters):
>
>'We all know that Mother Nature does not play the race card and did not
>intentionally pick a predominantly African-American coastline to wreck havoc.'
>
>It doesn't look like ADS-L has commented on this before.  The ghits stats are
>
>WREAK HAVOC     3,733,000 (including WREAKING / -ED / -S HAVOC)
>WRECK HAVOC     273,300   (including WRECKING / -ED / -S HAVOC)
>
>
How did you obtain these numbers?

If I search for "wreaked | wreaked | wreaking | wreaks havoc" (with the
quotation marks, on google.com, on whichever server I am sent to from
Europe), 100 results per page, skip to the last page (i.e. click on page
10), Google indicates a count of 1,630,000. This is still at least ~30%
too high (page 10 is in fact page 8, with only about 10 results: the
remaining 290 hits out of the 1000 Google displays at most are spam
pages, filled with words drawn from dictionaries and other pseudo-text;
their frequency can be expected to rise in the pages that aren't
displayed, since Google sorts by pertinence).

Similarly, "wreaked | wreaked | wreaking | wreaks havoc" gets you
136,000, minus 30%. (At least it's the same factor for both searches.
This is generally not the case.)

And then there's the, er, irregular past tense "wrought havoc": 60,400
minus ~35%.

Owing to the proliferation of fake web pages, Google counts -- and
worse, even relative Google counts -- are an extremely iffy business at
the moment. Yahoo! seems to filter more aggressively, but has its own
problems. And the search/filtering algorithms can change at any moment
without notice.

Thanks, Arnold, for posting the Eggcorn Database links.

Chris Waigl

--
blog:      http://serendipity.lascribe.net/
eggcorns:  http://eggcorns.lascribe.net/