A tipster informed us this morning that Twitter’s search engine (formerly known as Summize before it was acquired this summer) has been showing tweets in its results that don’t actually exist on Twitter itself.
These “phantom” tweets all contain the message “What a day 
. Wed, 08 Oct 2008″ plus a time in military format attached to the end. You can find them either by searching “What a day Oct” or simply “what a day ”. There are so many of these tweets, that the term “What a day Oct” showed up as a trending tweet on the homepage for awhile.
How do we know these tweets don’t actually exist? Just click through to any of the user accounts listed and you’ll see they don’t show up in the user’s archive.
It’s not clear what is causing these inaccurate results. If the search was somehow hacked, it would raise concerns about the security of Twitter’s API. If not, and these tweets actually originated on Twitter somehow (not out of the question, especially given how fake many of the affected accounts look), then it just goes to show how easy it is to spam Twitter Search.
Everyone, start panicking! Twitter API has been hacked!
This is big news. If al-Qaeda can do this, think of what else they might be capable of.
Just kidding. It was probably the PLO
A default search term, even if its due to a hack, makes it as a “Headline News” in TechCrunch.
Hmmm….I’m wondering if there will be a follow-up once we know why this was done by Twitter/Summize
They are not “phantom” because they did exist at one point. I posit that there is some rogue script running and posting to these accounts via the API, then immediately deleting the tweet via the API. The search.twitter.com database does not remove deleted tweets (something twitter says they are “working on fixing”), thus they appear in the search index. Some of the tweets actually appear on the users’ page if you click-through. It appears that most of the affected accounts are bots/spammers, but there are a few legit-looking accounts in the mix.
and why is this so significant. I guess we will find out later. Just implement stronger security.
What d day
.
OMG TECHCRUNCH GOT HACKED
“perhaps Summize has been hacked”… What a stupid, brainless thing to say. If I was hacking into Twitter the first thing I would do would definitely be to make a search return irrelevant results.
Maybe TechCrunch got hacked and they added this ridiculous article.
Today’s date also shows the same – http://search.t...h?q=08+Oct+2008
WTF it showed that i too posted the “What a day” tweet.. but nothing in my profile :-S http://twitter..../sizzler_chetan
Really? I don’t see any “what a day” from you in search: http://tinyurl.com/4hnos5
Seems like as far as Techcrunch is concerned, there is a dozen start-ups that are endlessly more important than thousands of others.
Last I checked, Twitter has just 2M accounts (and since these are free and many people have several, and many created an account and abandoned it, probably a lot less use it)…
Twitter is useless to me. I don’t understand the hype, at all.
ditto
same here
I think Twitter purposely crashes & messes up just to get more and more free PR like this (genius)
Duh, and yet fools continue to use it’s “service”.
Someone call DHS. This could be part of the pattern.
Im not big on twitter, you can use facebook status updates just as effectively.
YAWN…siesta time.
Twitter what now?
type webtrafficjuggernaut in google and you may find the answer…
I call it a bug. I’m willing to bet that there’s some fancy interpretation on the string “What a day” that the Ruby language is translating into a real date.
What do the Ruby folks out there think?
I think they fixed it already.
I just get frustrated because you can delete a Tweet but it lives forever in Twitter Searchland.
Those are signs from ET, you fools.
it’s still happening…
It may just be an easy way to spam twitter. I just gave it a shot. Add the tweet What a day
and then delete it. It sends you to the top of the search. Seeing as most of the search results are from fast cash sites, hairloss, etc. It seems it’s just spam.
Do not mess with twitter bird
There is an alternative answer to the “hacking” or “spamming” options. Perhaps this was some debugging code accidentally left in by a programmer? This would seem to make more sense to me as it is the same message appearing over and over. . .if I were a spammer, I’d be inserting links to a product. . .if I were a hacker, I’d be running a XSS attack or defaming the site.
Just a thought.
@Mark Nope, they are testing web-bots by having test accounts dump sample text. Next thing they will put the links to sites to drive traffic by putting random qoutes (like madlibes) i.e. “My laptop caught fire… I will buy a “
Well twitter APIis not robust, and I have noticed that when you delete a tweet twitter and search doesn’t remove the reference in the data base, so I would assume if you post a tweet every single minute and remove the tweet just after it is posted you generates phantom tweets which would appear in the search if they are numerous enough…
Speaking as someone who runs a search engine, all the wiseguys did was delete their tweets once they posted them. Adding messages to the index is easy, deleting them(as there’s no feed for that at least externally to Twitter) is another story.