i think the idea is that most of the time we ==don't want it to show up== 
because the signal-to-noise ratio is so crumby

if it were up to me, i'd toss most of the iK entries anyway.  iK should be used 
for common misrepresentations, not one-off henkan errors or "a handful of 
guys wrote it like this a few hundred years ago but so infrequently that writers 
of real Japanese dictionaries have never even thought of including it".  If 
writers consistently had to give yomigana so their readers could understand 
that they were using the reading じょうだん instead of しょうだん, that's a 
good sign that 笑談[じょうだん] was never in wide enough circulation to be 
relevant.

Replying to: "I think the point of putting in a comment is so that they don't get to be a headword. Even with an iK it creates something messy, e.g. for text glossing."

Would the mess come from the fact that there would be two entries with 笑談 as an headword ? 

I don't care about this that much, but it would make sense to me to include it as an headword since it's used to write じょうだん by a couple of different writers. 

The problem with the note is that if something is just written in a note, it won't show up when doing a search on wwwjdic. Of course, I doubt that many people will be searching for 笑談 on wwjdic, but you never know.

@@ -15,0 +15,4 @@
+</k_ele>
+<k_ele>
+<keb>笑談</keb>
+<ke_inf>&iK;</ke_inf>

nikk

Merging in Scott's suggested entry.

@@ -22,0 +22,10 @@
+<r_ele>
+<reb>ぎだん</reb>
+<re_restr>戯談</re_restr>
+<re_inf>&ok;</re_inf>
+</r_ele>
+<r_ele>
+<reb>げだん</reb>
+<re_restr>戯談</re_restr>
+<re_inf>&ok;</re_inf>
+</r_ele>
@@ -26,1 +36,1 @@
-<s_inf>sometimes written 笑談</s_inf>
+<s_inf>じょうだん is sometimes written 笑談</s_inf>
@@ -29,0 +39,1 @@
+<gloss>funny story</gloss>

@@ -25,0 +25,2 @@
+<xref type="see" seq="2589580">笑談・しょうだん・2</xref>
+<s_inf>sometimes written 笑談</s_inf>

I think the point of putting in a comment is so that they don't get to be a headword. Even with an iK it creates something messy, e.g. for text glossing.

Not that I want to stretch the debate here, but I prefer having Kanji headwords in the proper field and using the iK tag. I'm not a big fan of extra headwords hidden in notes and I think that the iK tag is there for situations like these.

The Soseki reference is nothing special and I see that it's used by others as well. (see http://www.google.co.jp/search?hl=ja&source=hp&q=笑談+じょうだん+site:www.aozora.gr.jp)

@@ -16,0 +16,4 @@
+<k_ele>
+<keb>笑談</keb>
+<ke_inf>&iK;</ke_inf>
+</k_ele>
@@ -26,1 +30,0 @@
-<s_inf>sometimes written 笑談</s_inf>

@@ -25,0 +25,2 @@
+<xref type="see" seq="2589580">笑談・しょうだん・2</xref>
+<s_inf>sometimes written 笑談</s_inf>

yeah, i'm not buying it.  笑談 should have its own entry.  and nikkoku lists its 
earliest usage as the 15th century, so it's not a soseki-ism

those aren't ateji, btw.

@@ -11,4 +11,0 @@
-<keb>笑談</keb>
-<ke_inf>&ateji;</ke_inf>
-</k_ele>
-<k_ele>
@@ -16,1 +12,0 @@
-<ke_inf>&ateji;</ke_inf>
@@ -20,1 +15,0 @@
-<ke_inf>&ateji;</ke_inf>
@@ -31,1 +25,0 @@
-<s_inf>笑談 is an ateji created by 夏目漱石</s_inf>

Rejecting this branch so I can approve the other.

Keeping it as a note without the headword would work for me.  I don't think they should be merged.

My inclination is to drop 笑談 from the entry apart from an "also written". It was added in Sep. 2008 by Scott with the comment about 夏目漱石. I can see comments in WWW pages about it being read じょうだん.
If it's not dropped, perhaps the entries should be merged?

@@ -31,0 +31,2 @@
+<xref type="see" seq="2589580">笑談・しょうだん・2</xref>
+<xref type="see" seq="2589580">笑談・しょうだん・2</xref>

nikk

笑談 is listed as しょうだん in both daij and koj so this might be a mistake.

@@ -12,0 +12,9 @@
+<ke_inf>&ateji;</ke_inf>
+</k_ele>
+<k_ele>
+<keb>戯談</keb>
+<ke_inf>&ateji;</ke_inf>
+</k_ele>
+<k_ele>
+<keb>串戯</keb>
+<ke_inf>&ateji;</ke_inf>