This approach becomes difficult if you need to enforce domain constraints within the database through the use of referential integrity.  Such a solution forces then constraints into the program code rather than in the database.   In a complex enterprise application this leads to poor data quality.
It also adds complexity if you have very complex reporting requirements .

You can also support multiple languages with this scheme by adding a LanguageID column and filtering on it (each distinct lookup item having the same LookGID but appearing with different LanguageID values).  Of course, if you are not planning to translate everything, you then need a mechanism to pull out some default value (say English) when items are not found.

I am not going with the author or the mainstream here.

As I'm too tired to write a pamphlet and in addition have only a very limited english vocabulary compared to the author of the reply in the link below, read this

http://groups.yahoo.com/group/sql_server7_forum/message/5731 

It's a reply to this question

http://groups.yahoo.com/group/sql_server7_forum/message/5726

Use of a cursor for calculation of the median (financial or statistical ?) ???

CREATE TABLE #dist (c1 int)
INSERT INTO #dist VALUES (2)
INSERT INTO #dist VALUES (3)
INSERT INTO #dist VALUES (1)
INSERT INTO #dist VALUES (4)
INSERT INTO #dist VALUES (8)
INSERT INTO #dist VALUES (9)

SELECT Median=CASE COUNT(*)%2
   WHEN 0 THEN
    (d.c1+MIN(CASE WHEN i.c1>d.c1 THEN i.c1 ELSE NULL END))/2.0
   ELSE d.c1 END  -- Odd number
FROM #dist d CROSS JOIN #dist i
GROUP BY d.c1
HAVING COUNT(CASE WHEN i.c1 <= d.c1 THEN 1 ELSE NULL END)=(COUNT(*)+1)/2

DROP TABLE #dist

Median             
-------------------
3.500000

or if there are duplicate values

set nocount on
CREATE TABLE #dist (c1 int)
INSERT #dist VALUES (2)
INSERT #dist VALUES (2)
INSERT #dist VALUES (1)
INSERT #dist VALUES (5)
INSERT #dist VALUES (5)
INSERT #dist VALUES (9)

SELECT Median=ISNULL((CASE WHEN COUNT(CASE WHEN i.c1<=d.c1 THEN 1 ELSE NULL END) > (COUNT(*)+1)/2 THEN 1.0*d.c1 ELSE NULL END)+COUNT(*)%2,
              (d.c1+MIN((CASE WHEN i.c1>d.c1 THEN i.c1 ELSE NULL END)))/2.0)
FROM #dist d CROSS JOIN #dist i
GROUP BY d.c1
HAVING (COUNT(CASE WHEN i.c1 <= d.c1 THEN 1 ELSE NULL END)>=(COUNT(*)+1)/2)
AND (COUNT(CASE WHEN i.c1 >=d.c1 THEN 1 ELSE NULL END) >= COUNT(*)/2+1)
GO
drop table #dist

Median          
----------------
3.5

There are many other ways possible. None of them utillizes a cursor.