cc licensed flickr photo shared by Mizrak
While Twitter is all about “what are you doing now” it is mostly useless to try and find out “what I did last month”. The search capability feels more like pulling a slot machine handle in terms of feeling like you will hit a jackpot.
The Library of Congress cannot move fast enough for me.
Here is my case example. In an email exchange at work, the nifty web app notebook.cc came up. I knew I had played around with it sometime recently. I could not find a blog post here, so I went to the twitter search (advanced searched on tweets by me and “notepad.cc”)
However, I did not buy those results. I knew I tweeted it, damn you finicky bird!
So next step? Go to the Oracle. I googled on @cogdog notepad.cc and found mentions, references to it, and eventually got there because luckily @karlyb had favorited it.
So here you go, lame a** twitter search bird here it is July 16, 2010, and you are not able to find something on your own servers, that is less than 30 days old??
Yep, it is right there http://twitter.com/cogdog/status/16630806268.
Twitter search therefore is not really a search- I do not even know what to call something that does not even index its own content in a manner that is close to comprehensive. It might be interesting to do some experiments to find out how big these gaps are.
So what can you do? First, if you tweet something you may want to recall, maybe the answer is to favorite it. Google is always an option, as it seems to be indexing twitter in some amount (maybe twitter is hoping for the Google-Buy-Me dream?)
I’m going one better, I just set up a twapper keeper @person archive, which archives your own tweets — http://twapperkeeper.com/person/cogdog. This service is becoming more and more powerful for my work; I’m looking next at making an archive of archives for our NMC content. Again its case of a third party well executing something the primary party is not doing. That is one gain my the primary party (twitter) at least providing the APIs for others to fill their gaps.
I also have backupify running on my twitter account, but have yet to go peek in there.
Got any other solutions to better track your tweets?
But wake up, Twitter Search is not something you want to ride in to the prom
cc licensed flickr photo shared by thorinside
And i should add that if you use a twitter widget to put the tweets for your event tag on your web site, you can count on the stream going dry a few weeks after. My current strategy for events is to (1) create a twapper keeper archive ahead of time; (2) use a twitter widget during and a few weeks after the event; (3) later replace the widget with a link to the twapperkeeper archive.
As a sideline, I find that Google gives a lot of false hits on my blog, because I have… er had.. been using the twittertools plugin widget to insert my tweets on the sidebar. What this means is that Google is indexing content (the tweet stream) at the time of indexing that has nothing to do with the content on my blog post. I slapped that widget off the screen.
You should try to find a way to voice this to the LC. When one of their team spoke at CHNM right before THATCamp, she talked about not really being interested in the kinds of architecture that would allow people to do the kinds of searches, because the LC wants to be very careful not to reduplicate functionalities already built into twitter.
Which makes sense, in a way, you want to be both more and less than a mirror site, I get that, but at the same time– if even a couple months in the past is lost to twitter due to poor search, that’s relevant to the types of things they should be looking at trying to do with the archive.
Was experiencing exactly the same problem and like you now have multiple ways to backup tweets, one of them being a wordpress plugin I wrote (twitterpad) which collects tweets http://www.rsc-ne-scotland.org.uk/mashe/tweets/ and puts them on a page. Whilst better solutions have come along I still use the plugin because I have full control off the backups (and gives my blog more exposure on search indexing)
I’m doing a brute force capture of all my tweets to an Open Office text file.
While not a great solution you can link your evernote account to twitter and add @myen to save tweets directly to your ever note.
This is true. Older Tweets aren’t indexed, or don’t seem to be. As you’ve pointed out, a third-party search service built around the Twitter API, however, can retain those older Tweets, at the very least around a topical slice of Twitter. Essentially, that’s what we’ve been doing, filtering Twitter’s stream for just the well-formed questions and making that a real-time searchable stream (http://replyz.com/c.) Because we don’t remove older questions, at least those with responses, from the index, we’re able to build a growing archive of question-Tweets and their @replies that’s deeper than Twitter search. So, it’s certainly possible for niche search services built around the Twitter API to produce arbitrarily deep archives of Tweets in some cases. Haven’t surveyed all the Twitter search apps out there, but I imagine others are doing this, will have to take a look.