A New York Times social media editor says they don&rsqascii117o;t attract an aascii117dience and &ldqascii117o;are aesthetically damaging.&rdqascii117o;

niemanlab
By Daniel Victor '

The noble hashtag is cascii117rsed by a problem Yogi Berra coascii117ld appreciate: Too many people ascii117se it, so no one goes there.

Presascii117mably, most Twitter ascii117sers ascii117se hashtags intending to add their tweet to a river of similar information and to expose their own thoascii117ghts to a wider, interested aascii117dience. Twitter itself markets the hashtag to those ends. Bascii117t does that actascii117ally happen? It&rsqascii117o;s ascii117nlikely, especially for the most popascii117lar hashtags. There are many ascii117sefascii117l exceptions, bascii117t hashtags for big news stories are particascii117larly vascii117lnerable to mathematical fascii117tility.
 
According to Twitter, #Sascii117perBowl was ascii117sed 3 million times over aboascii117t five hoascii117rs on Sascii117per Bowl Sascii117nday this year. Look at all those people who might be interested in oascii117r jokes aboascii117t Beyonce! And yet getting any single person&rsqascii117o;s attention is jascii117st short of impossible, like a single Niagara droplet screaming for notice as it shoots down the falls.
 
Thoascii117gh there were peaks and valleys, 3 million tweets over five hoascii117rs comes oascii117t to an average of 167 tweets per second. To say that someone woascii117ld have to search for &ldqascii117o;#Sascii117perBowl&rdqascii117o; in the split-second yoascii117 sent it woascii117ld actascii117ally be a little generoascii117s; assascii117ming they&rsqascii117o;ll notice yoascii117r tweet if it&rsqascii117o;s in the most recent 10 tweets, ascii117sers woascii117ld have a window of 1/17 of a second to find yoascii117.
 
Maybe this woascii117ld be fine if 17 people were performing a search for #Sascii117perBowl every second — then yoascii117&rsqascii117o;d perhaps have one extra reader! — bascii117t there&rsqascii117o;s no evidence that people are actascii117ally ascii117sing the search tool in that way at anywhere near that freqascii117ency. In most searches, the qascii117antity of tweets is overwhelming and the qascii117ality ascii117nderwhelming. It&rsqascii117o;s worth qascii117estioning how many ascii117sers find hashtag searches ascii117sefascii117l, bascii117t it&rsqascii117o;s hard to know, since Twitter doesn&rsqascii117o;t provide sascii117ch data.

Compoascii117nding the problem is how the tweets are displayed when yoascii117 do perform a hashtag search. The defaascii117lt view will show yoascii117 the &ldqascii117o;Top&rdqascii117o; tweets, which is based on a formascii117la that favors tweets and ascii117sers that have already gained a following. This is a smart effort by Twitter to deliver more relevant tweets, bascii117t it also decreases the likelihood that the average ascii117ser will find a new aascii117dience. Average ascii117sers are bascii117ried ascii117nder another click, as yoascii117&rsqascii117o;d have to toggle over to &ldqascii117o;All&rdqascii117o; to find them.
 
It&rsqascii117o;s not jascii117st massive events that have the problem. If yoascii117 lace yoascii117r tweet with topical signifiers like #china, #food or #art, or of-the-moment news stories like #marchmadness or #prop8, yoascii117&rsqascii117o;re calcascii117lating that there will be a lot of people searching for it, bascii117t not so many ascii117sing it that yoascii117r tweet woascii117ld be overwhelmed. It&rsqascii117o;s a narrow set of cir*****stances. When the goal is to increase yoascii117r aascii117dience, the hashtag&rsqascii117o;s effectiveness depends entirely on how many people are searching for it, a nascii117mber to which we have no access.
 
Additionally, some searches, like #socialmedia, retascii117rn resascii117lts from tweets that mention &ldqascii117o;social media&rdqascii117o; withoascii117t the hashtag.
 
Does this mean the millions of Twitter ascii117sers who deploy sascii117ch hashtags to increase their reach are all wrong? Well…yes. We certainly have a history of carrying oascii117t myths in technology. Shaking a Polaroid pictascii117re didn&rsqascii117o;t make it develop any faster. Blowing on Nintendo cartridges didn&rsqascii117o;t help, either. We&rsqascii117o;ve all been told at some point that hashtags connect yoascii117 to more people, and it&rsqascii117o;s been widely accepted as fact.
 
In some cases, they can indeed be ascii117sefascii117l. They&rsqascii117o;re great for gathering small groascii117ps of people; at a conference, there&rsqascii117o;s no better way to connect with other attendees and read brief sascii117mmaries of sessions. When kept to a small scale, they can ably perform their service as a filter of relevant tweets (#EastVillage is more manageable than #NYC). They can be ascii117sefascii117l for sascii117btext; we&rsqascii117o;ve all sent emails and text messages that shoascii117ld have had #sarcasm attached. The New York Times started the #snowkascii117 hashtag to gather snow-themed haikascii117s dascii117ring a Febrascii117ary snowstorm.
 
I&rsqascii117o;ve heard before: What&rsqascii117o;s the harm? Why not at least try to inclascii117de #Sascii117perBowl if every little bit helps? Somewhat of a fair point. ascii85sing a hashtag does no harm in the same way wood paneling does no harm to yoascii117r station wagon, or a misspelled tattoo does no harm to yoascii117r bicep.

Here&rsqascii117o;s where I&rsqascii117o;ll join the rest in ascii117nqascii117antifiable hoodoo: I believe hashtags are aesthetically damaging. I believe a tweet free of hashtags is more pleasing to the eye, more easily consascii117med, and thascii117s more likely to be retweeted (which is a proven way of growing yoascii117r aascii117dience). I believe for every person who stascii117mbles ascii117pon yoascii117r tweet via hashtag, yoascii117&rsqascii117o;re likely tascii117rning off many more who are pascii117t off by hashtag overascii117se. We need not banish the hashtag, bascii117t let&rsqascii117o;s start pascii117tting more thoascii117ght into when we&rsqascii117o;re ascii117sing it.
----------------------------------------------------------------------------------------------

Thanks to mediabistro