Twitter Search Archives

I’m not sure if this feature is available to everybody yet, or whether I’m just very late to the party, but earlier this morning when using Twitter search I noticed something new at the top of the results – Recent Retweets.

It doesn’t seem to work for all keywords, but on certain occasions Twitter has started ranking certain high-profile users above the normal reverse-chronological results that are normally generated.

For example, here is a search for Twitter:

Twitter Adds ‘Recent Retweets’ Feature To Search – Is This The Start Of Reputation Ranking?

Read the rest of this entry

Over at the Twitter API announcements Google group, Twitter developer advocate Taylor Singletary has written about a beta project that their search team is working on that will rank results by popularity (as opposed to reverse-chronologically as they are now).

The Search team is working on a beta project that returns the most popular tweets for a query, rather than only the most recent tweets. This is a beta project, but an important first step to surface the most popular tweets for users searching Twitter.

You can expect many improvements as we tune and tweak our algorithms, but we want to give everyone a heads up so we can go over the implications for those consuming the search API.

It’s unclear how exactly they are going to define what it is that classes a tweet as popular. People rarely click on individual tweets, so that metric isn’t going to work at all. My guess is it’s likely to be based heavily on retweets, perhaps held up against network size, which means that Justin Bieber is going to be your authority on everything.

Most Popular Tweets Soon To Rank First In Twitter Search

For everybody’s sake, let’s hope they’re also working on an algorithm to counter that, too.

(Hat tip to Mashable.)

Twitter plans a Google Adwords-style advertising model, according to All Things Digital.

(Read more at 140char.com.)

Ads will be delivered via searches on Twitter, and come packaged in 140 characters or less, which might present a dilemma for businesses to get their message across. That said, we’ve all had a lot of practice at selling tweets, so advertisers should be at least semi-prepared.

(Imagine how much better this would all be if Twitter searches came with TweetRank? Perhaps the users with the most clout could be linked up to the advertising model and rewarded accordingly.)

This is a bit of a no-brainer for Twitter. I’ve often speculated on the plausibility of an Adwords-style system on the network (using Spotify as a case study) and the most surprising part is that it’s taken this long to implement (and still won’t hit the platform for a few months yet).

It’s really too early to speculate about the consequences of all of this without more information, but I do have one question: will the option to advertise be open to everybody, like it is with Google’s Adsense program, or is Twitter going to continue the form it has shown with the suggested user list and verified users, and only offer this service to their personal favourites?

When calculating rankings, Google often penalises websites for doing various things, including cloaking, automated queries, irrelevant keyword use, paid links, sneaky URL redirection, malicious behaviour and linking to known spammers.

I wonder if Twitter needs to embrace a variation of Google’s PageRank system – let’s call it TweetRank – and begin to score and rate individual accounts according to the ways in which they behave, handing out penalties to users who show a blatant and repeated disregard for the rules.

Cloaking

Cloaking is the practice of deception by displaying different content to search engines than that which is displayed to users. There is a frighteningly large number of Twitter users who act very above board and ‘normal’ in the public side of the network, while functioning as little more than mass-marketers and spammers in the private world of direct messaging. Direct messages are private, and none of us like the idea of ‘the man’ reading our inbox, but a warning system could work well here, where users flag accounts for duplicitous behaviour.

Automated Queries

Google doesn’t like it when websites bombard them with automated queries, as it wastes resources and bogs down server time. Automated messages on Twitter – which includes direct messaging and updates from external services (including things like Foursquare) – are equally undesirable. Often many users are unaware that they’re sending out automated messages, or that their accounts have been exploited in some way, because they don’t regularly check their connections settings.

Irrelevant Keyword Use

Some users include words in their bios that are clearly there simply to generate results from searches. Common examples are SEO, make money at home and profit. In many cases these accounts are nothing more than spam feeds that heavily-promote affiliate schemes and the like.

Paid Links

Advertising on Twitter is a hot potato. From Twitter’s perspective, it’s an inevitability. But what about users that get paid to link to external sites? Advertising for publishers is very much part of the internet’s business model, but if these paid links go out to scams or make claims that are unproven or entirely false, then the user should take responsibility and be marked down accordingly.

URL Redirection

It’s a common scam to hide bad links within good ones. I’ve seen some users hide an affiliate or malicious link between several layers of bitlys and TinyURLs. It is my personal opinion that any user that links to a malicious website or known scam should be heavily punished by Twitter, perhaps adopting a ‘three strikes’ rule to avoid accidental retweeting or exploits.

Linking To Known Spammers

This is perhaps the most important item on the list. Despite their best efforts, Twitter continues to have major problems with spam. This issue is made significantly worse by people who blindly auto-follow anybody who follows them, because it legitimises the spam account, both in terms of improving their ratio and showing an ‘A-list name’ in their network. Twitter needs to take greater responsibility in ridding the network of obvious spam accounts, but the users need to step up, too, and I propose that any user who follows too many spammers should be punished.

(This would also allow the implementation of a reward system that hands out TweetRank bonuses to users who are followed by other highly-ranked individuals, providing a greater indication of clout.)

What Kind Of Punishment?

Users who breach some or all of the rules above are penalised, with a lowering overall score reflecting how highly you place on Twitter search. Score a ten, and you show up for all relevant queries, right at the top of the list.

Score a zero, and you don’t show up at all. Ever.

This could make Twitter search an incredibly powerful and reliable system, as results could then be measured by clout and reputation, as well as ridding the mechanism of the bulk of spam and scammy or malicious links. Twitter could add a little relevance option to the search results that re-ranked the output according to status, or you could view the data reverse-chronologically, as we do now.

Perhaps – and this is controversial – all new users to Twitter should have to ‘earn’ their place on the search results, much like all new websites have to earn their spot at the top of Google. Your TweetRank rises and falls with your behaviour. Some people won’t care too much about where they place in Twitter’s search results, but for businesses, brands and influencers it’s incredibly important. And as the network expands, the value of ranking well on Twitter search will rise exponentially.

Moreover, accounts that start spamming right out of the gates will be hidden from everybody else, and likely will never earn a spot within search.

Once users are ranked, they can easily be tagged and categorised, and finding the top 100 experts on any given subject would be available to you at the click of a button or two.

Rewarding and penalising users is potentially a risky endeavour and, much like Google, Twitter would need to keep revising and revamping the algorithm that they would use for any TweetRank system. An independent ombudsman could be created to ensure fairness, and perhaps the opinion of the Twitter collective could be a factor in a user’s score. There would inevitability be a teething period, and a strong likelihood of cries of foul play and favouritism from some quarters, but the benefits to the network as a whole should not be underestimated.

Yesterday, Twitter rolled out its Local Trends feature to everybody on the network.

When you login to Twitter.com, you’ll be given the opportunity to set the trending topics to your choice of 22 different locations.

Local Trends Is Only Slightly Less Useless To Me Than Global Trends, Twitter. Why Can’t I Search Just My Network?

Immediately, of course, you see the problem here – while Twitter has stated that they’re working on adding new locations, chances are that most people who use the service will find that Local Trends is anything but for them at this moment in time.

Indeed, the nearest ‘local’ to me would be London. And, being absolutely frank, being able to quickly see what’s trending on Twitter within London isn’t of much more benefit to me than being able to quickly see what’s trending on Twitter everywhere.

Moreover, the differences between what is trending around the world, in the United Kingdom and within London aren’t as staggeringly different (or interesting) as you might expect.

Local Trends Is Only Slightly Less Useless To Me Than Global Trends, Twitter. Why Can’t I Search Just My Network?

As you can see, the UK are clearly more interested in the tennis at the Australian Open than the rest of the world, and you would expect a UK-specific event such as Holocaust Memorial Day to only be trending within these shores.

And it amuses me that Londoners seem to favour the iPad as the name of choice for Apple’s soon-to-be-announced touchpad device, while everybody else is all about the (God-awful) iSlate.

Otherwise, it’s all about the glory of #nowthatsghetto.

Here’s the thing: being able to see trending topics in any given location is always going to be less useful to me than the facility to be able to set whatever filters and parameters I like from scratch.

For example, the option to simply search within the tweets of those people in my immediate network on Twitter would be something of significantly higher value. Using this method one could quickly and easily poll trends and opinion from those whose judgement we already trust. (And if not, why exactly are you following them?)

And being able to remove trending topics that are of no interest with the touch of a button would also improve the output dramatically. I accept that hashtag memes are popular with many, but they’re of absolutely no interest to me whatsoever. I’d rather not see them, to be honest, if only because they’re taking up space that might be better used for something that’s actually of value.

As it is – and I dare say this would be the case if Twitter ever gets round to providing local trends for my home town – aside from the novelty value, seeing what is trending in any one location is never going to be of much benefit. If something big happens in London and is only trending in London, that’s really only of use to Londoners. And even that’s a stretch, as if it’s big enough to be trending, it’s likely already in the greater consciousness, as that’s why it’s trending.

(Although how or why Danny Dyer’s television show about UFOs has generated such a level of interest is anybody’s guess.)

And that’s the rub in a nutshell: the problem with trending topics is that they are too general. This is why they’re trending, of course, because lots of people are talking about them, but while you can occasionally get surprised by something on a trending topic, when it’s still trending a week later it’s little more than an irritant. Especially when the reason why it’s trending is wrong.

If we could shape this output the value would increase exponentially. And it would be nice to see outside just the top ten, too – after all, chances are that there are many worthy things of interest in the long tail between numbers eleven and infinity, but the bulk of these never get a look in simply because once a trending topic is inside the top ten it automatically gets a lot more attention. In this way, it works a bit like a music, movie or book chart. The real quality is often just outside the very top, going unnoticed by the general populace.

It’s good to see Twitter rolling out all these new features of late but I do wish they’d adopt a sense of priority. We still have major problems with Twitter search, a block function that doesn’t work, a very ropey direct message system and never-ending problems with spammers and bots.

Perhaps if and when these things begin trending we might start to see some solutions. Of course, if you’ve set your Local Trends to the wrong location, chances are you won’t even notice.