Is gmail being down a fail? After all it's only a beta product...
catarino™
· 7 months ago
#goats_chew_gmail_cables :)
Sheamus
· 7 months ago
Even a polished reputation doesn't mean one fails to make mistakes, even in one's own niche. Yesterday, Robert Scoble announced that Wolfram Alpha was now live; cue, tens of thousands of disappointed people.
Matt Cutts is a good example of where reputation rank would work regarding Google, but that's a very simplistic case. What about information that is more art than science, more philisophical than factual? Would Richard Dawkins have a higher reputation on religious matters than the Archbishop of Canterbury?
Even something like the trust in reportage is very much in the eye of the beholder; one man's BBC is another man's Fox News.
I suppose my question is: who decides on the reputation? Twitter isn't all about technology and social media, despite how it may seem, and as the user-base expands this will become increasingly obvious. It might be nice if we were allowed to apply our own parameters about whose reputation we value.
guruvan
· 7 months ago
the unfortunate thing of this is that twitter is going to likely be the ones who decide on it, and that about guarantees that the reputation system is going to be skewed away from the non-celebrity experts on various subject. and if this is the case, this will certainly diminish the value of twitter's search in many eyes. Then other methods of searching through the volumes of not just tweets, but dents, and bellows, and friendfeed posts and comments, will be devised, with their own version of reputation scores.
henrymack
· 7 months ago
We've been using a contextual relevancy algorithm to rank job tweets in addition to indexing URLs for http://TwitterJobSearch.com
adityarao310
· 7 months ago
Well, I think all this talk is great. But Twitter is also planning to index the links in tweets and the contents in those links.
1. Almost all links in tweets are in short form, wont this affect the indexing process? 2. Can twitter be expected to come out with such drastic and active steps when they haven't been able to come up with even a decent business model?
Just a counterview. I love the new idea behind Twitter search, personally :)
guruvan
· 7 months ago
1) it's trivial to expand the links 2) twitter is using bit.ly which they have a relationship with 3) it's expected that twitter will acquire bit.ly 4) business model, hell they can't get their core business working right, and you expect someone to pay for it ;-) lol.
the money is actually in the search, but I really do think they need to get the architecture to a point where it will scale before search is really something to worry about.
Andrew Webb
· 7 months ago
I definitely think reputation rank is important for Twitter search. I have summarised my thoughts on what i call TwitterRank in my blog on ideas on how to improve Twitter search - http://openenterprise.wordpress.com
Johnny
· 7 months ago
Stan, you make some great points. However, it is past time to put pressure on Google to provide real time information on network issues and be responsive to its users. I am having an ongoing problem with Google Reader ( RSS ) which I have totally integrated into my daily business routine for over a year. In frustration, I called Google HQ in Mountainview, Ca. and the switchboard operator told me that Google does not provide any live support for any of its products. When I insisted that she connect me to the Google Reader development group she blocked my phone number. This kind of high handed approach to dealing with competent end-users is unconscionable given the amount of money Google rakes in from our collective use.
Mark H. Delfs
· 7 months ago
How about suggesting to use Google Gears--which in theory, would not make a overly huge difference if Gmail went down for a few minutes here or there--granted, you would not get new mail, but, you would have all the access to your current mail. Even with a 20 minute outage here or there, I'm not moving away from any of the Google products for anything.
MrGoogleAlerts
· 7 months ago
Twitter has the best possible development team for solving this problem: the thousands of startups that think they are going to get rich off of the Twitter stream. Twitter's internal development team should focus on managing the Twitter load, which still overwhelms it on a daily basis. When a brilliant solution for real-time reputation ranking emerges from the pool of Twitter add-ons, Twitter should buy it. That is the most cost effective plan, and the one that is most likely to succeed. The fact that Twitter's internal team thinks they have the intelligence to solve such a complex problem at the same time they are trying to keep the site from failing proves that they don't.
A twenty minute outage on a busy trading day is indeed a big blow. Not great for customer service. However, we have to recognise that Gmail has been going 5 years and I have never paid a penny for it, and 99.999% of the time it has offered an excellent service. Indeed, Google has changed our lives in a very positive way. Rgds Vince
FadiPick
· 7 months ago
Exactly! I love my gmail, it is very fast and much better than what other companies offer. I have a yahoo mail account and a hotmail one along with my google account. The Gmail is far way better. It is much faster and easier to use.
I can understand such outage if they are not so frequent, and they are not.
guruvan
· 7 months ago
you shouldn't have to pay for a beta product. ;-)
Justin Parks
· 7 months ago
While im confident that twitter will figure the twitter search out effectively I will be severely disappointed if they base the search solely on followers and RTs, a far to simplistic model for producing effective results in any form of search and open in more ways than one to a mountain of useless spam and drivel.
Twitter are stepping into the world of search and will have to produce an algorithm that is as complex as Googles in order to make it one thing and in a word that's "useful". It will fail whale totally if tweeple simply cannot find relevant results, simple as that.
To reply to Aditya Rao: 2. Can twitter be expected to come out with such drastic and active steps when they haven't been able to come up with even a decent business model?
I cant remember Google having a business model in the early days either. Now that search is of such obvious importance, and all the differing revenue streams associated with it, Google effectively created their own business model, the main barrier was users.
Get the users and the model can be whatever compliments the service and attracts more users. I think Twitter are following the same line and Search is their business model!
Addressing Gmail being down. I dont care, its free, its up enough to be damn useful, and frankly I don't think I have a right to complain when someone has handed me a service like this, no ones perfect and s**t happens to the best of us. (but still, communication is key and a simple announcement to users would sort most of their problems out, I thought they would be more than aware of this.)
Dawn
· 7 months ago
I have to totally agree with your assessment of the twitter ranking system, there has to be more to than someone's "status" within twitter.
Neal
· 7 months ago
Twitter's achillies heal is the almost total lack of metadata around twitts. All you get is a statement and you have to figure based on what is going on around you and through a rough search tool what the context of that statement really means.
FriendFeed does a better job of adding value thought comments and likes, but they haven't been able to figure you how to capitalize on that yet.
Nickd
· 7 months ago
You are quite right that measuring the reputation of social news (Digg) or social media (Twitter) submitters is complex problem. There are (at least) two fundamentally different approaches:
First is to highly constrain the input side (i.e. ask structured questions and push content into categories from the start, and/or require independent verification of submitter credentials). We ( http://www.vanno.com/ ) use the former, while Wikipedia is a good example of someone using the latter). The advantage is that the noise is cut down from the start. The disadvantage is the user/submitter base is much smaller, and submitters have to do more "work".
The second approach is to try and infer user reputation (credibility) of unverified submitters from their unstructured activity stream. This is what Twitter has to do. It's a horrible signal to noise problem, made worse by the fact that the target is always moving - i.e. people will always try and game any system or reverse engineer any algorithm. The advantage of this approach, of course, is that you can get an enormous user base and activity stream quickly, and submitters don't have to think much before hitting the send button.
Our bet, for what it's worth, is that the user reputation backward inference problem will ultimately prove intractable for completely unstructured activity streams - i.e. the defense (prevent gaming) will be overwhelmed by the offense (people trying to game the system). What's going on with Digg's attempts to keep up with systematic and professional gaming of user "reputation" support this view.
We're betting on the structured input approach - i.e. suppress the noise from the start. We wish Twitter good luck, but bet that in the end they'll be forced to expand hashtags to more structured categories, and enforce some level of a priori submitter "certification".
I think it would be a colossal brand-ending error for Twitter to introduce a reputation ranking system.
Even the best system I can imagine (a second-by-second ranking of each tweet based on each person's independent assessment of the value of the tweet) would work against Twitter's central idea: a foundation of egalitarianism, the efforts of people to stand out from the noise, the efforts of people to pan-handle the noise for good information, and an element of luck.
Reputation with respect to some specific tweet is very much a contextual issue and what may have warranted high ranking even a moment ago may no longer warrant it now for a whole slew of reasons. I do no look forward to the day when my stream is overwhelmed by highly ranked tweets all saying the same thing because those "in the know" are only talking to each other. I could go to the dailies for that.
I think the best method for this assessment of the value of a tweet is the ever alert folks who are doing that right now and pan-handling the stream quite effectively. For example, if I know someone who should know something on some issue I seek out their tweets -- I don't need an algorithm to decide for me. A better search tool would help, but based on content not reputation.
mark harrison
· 7 months ago
I think the idea of ranking Twitter users or their tweets stinks. All that will happen is that the 'gurus' will take over. Great news.
bravelittlememe
· 7 months ago
Tumblr has just launched a similar concept, "Tumblarity." I'm still not sure any of these ranking systems do anything but self-perpetuate already popular sources of information: the most popular sources become even more visible, thus becoming more popular, and so on. This ends up making it nearly impossible for the internet status-quo to be upset in favour of more relevant, but less known, content authors.
Stefan aka @susuh
· 7 months ago
Being on the "hand-picked link sharing side of Twitterers (that group around social media for small biz in general)" I naturally like the idea of reputation and (keyword based) relevance, e.g. as described in Andrew Webb's blog. (comment #3)
This kind of search will turn Twitter quickly into the preferred search-engine for all "fresh" information – and maybe even make bookmark services redundant in the long run...
Preternatural
· 7 months ago
One way Google could compensate for Twitter's "realtimeishness" would be to index and rank search queries, which are realtime. Google being at the top of the search heap, most people instinctively go right to Google to check up on something that's going on right now, though more out of habit than actually getting something of use. That said, its safe to assume that any time there is significant chatter on Twitter about a topic, there may be a corresponding (not necessarily equal) volley of searches about the same topic on Google concurrently.
Whitney
· 7 months ago
I certainly think retweets can be a great way to spread information within your network, but it's like playing postoffice- it's merely repeating information, rather than creating it on your own. If your reputation is going to be ranked by whether or not anyone ever retweets information, we could all easily end up with a twitter stream full of nothing other than other people's regurgitated ideas. Relevancy is more than whether or not something is passed along the chain.
Jared O'Toole
· 7 months ago
I like the idea. However I usually don't have a problem finding information. Someone big on twitter might have said it or a nobody but its still in the search and relevant.
Ari
· 7 months ago
I am incredibly excited by what I call, the multi-layering of social media content. Google's Maps and Street View integration with Twitter posts (WHERE), trending, and in this case reputation are all a part of understanding what is popular. But, the exciting part about appending data to the social media packet (the WHAT), is the ability to begin to understand the WHY. I think that social media should begin to aggregate more user data through the use of cross-platform web apps like feed.ly and bit.ly (the ly's, lol). Because of commonly associated aspects of a user (screen names, emails, etc.) companies can earmark individuals and collect information about them from across many social media platforms (Twitter, facebook, G-platforms). As this becomes associated, and new packet data gets appended (device, browser, screen settings, frequency), we can begin to collect information to explain more of the WHO, the HOW, the WHEN, and I believe, through algorithmic "stuff" the WHY of the social media trend could become clear... It requires a lot of data and types of data to be included in the model, but I am excited for the day when social media will begin to predict our needs, and inform us properly.
Smart Boy Designs
· 3 months ago
I think a ranking system would be quite helpful. Those who are more knowledgeable about a certain keyword should definately rank higher.
tobylane
· 3 months ago
It's moving away from the REALLY CRAP two dimensional method they are on. I wish they'd have the guts to give us something sane like 180 or 200 characters, rather than some bs about people we don't care enough to follow.
All Comments
Matt Cutts is a good example of where reputation rank would work regarding Google, but that's a very simplistic case. What about information that is more art than science, more philisophical than factual? Would Richard Dawkins have a higher reputation on religious matters than the Archbishop of Canterbury?
Even something like the trust in reportage is very much in the eye of the beholder; one man's BBC is another man's Fox News.
I suppose my question is: who decides on the reputation? Twitter isn't all about technology and social media, despite how it may seem, and as the user-base expands this will become increasingly obvious. It might be nice if we were allowed to apply our own parameters about whose reputation we value.
1. Almost all links in tweets are in short form, wont this affect the indexing process?
2. Can twitter be expected to come out with such drastic and active steps when they haven't been able to come up with even a decent business model?
Just a counterview. I love the new idea behind Twitter search, personally :)
2) twitter is using bit.ly which they have a relationship with
3) it's expected that twitter will acquire bit.ly
4) business model, hell they can't get their core business working right, and you expect someone to pay for it ;-) lol.
the money is actually in the search, but I really do think they need to get the architecture to a point where it will scale before search is really something to worry about.
I can understand such outage if they are not so frequent, and they are not.
Twitter are stepping into the world of search and will have to produce an algorithm that is as complex as Googles in order to make it one thing and in a word that's "useful". It will fail whale totally if tweeple simply cannot find relevant results, simple as that.
To reply to Aditya Rao:
2. Can twitter be expected to come out with such drastic and active steps when they haven't been able to come up with even a decent business model?
I cant remember Google having a business model in the early days either. Now that search is of such obvious importance, and all the differing revenue streams associated with it, Google effectively created their own business model, the main barrier was users.
Get the users and the model can be whatever compliments the service and attracts more users. I think Twitter are following the same line and Search is their business model!
Addressing Gmail being down. I dont care, its free, its up enough to be damn useful, and frankly I don't think I have a right to complain when someone has handed me a service like this, no ones perfect and s**t happens to the best of us. (but still, communication is key and a simple announcement to users would sort most of their problems out, I thought they would be more than aware of this.)
FriendFeed does a better job of adding value thought comments and likes, but they haven't been able to figure you how to capitalize on that yet.
First is to highly constrain the input side (i.e. ask structured questions and push content into categories from the start, and/or require independent verification of submitter credentials). We ( http://www.vanno.com/ ) use the former, while Wikipedia is a good example of someone using the latter). The advantage is that the noise is cut down from the start. The disadvantage is the user/submitter base is much smaller, and submitters have to do more "work".
The second approach is to try and infer user reputation (credibility) of unverified submitters from their unstructured activity stream. This is what Twitter has to do. It's a horrible signal to noise problem, made worse by the fact that the target is always moving - i.e. people will always try and game any system or reverse engineer any algorithm. The advantage of this approach, of course, is that you can get an enormous user base and activity stream quickly, and submitters don't have to think much before hitting the send button.
Our bet, for what it's worth, is that the user reputation backward inference problem will ultimately prove intractable for completely unstructured activity streams - i.e. the defense (prevent gaming) will be overwhelmed by the offense (people trying to game the system). What's going on with Digg's attempts to keep up with systematic and professional gaming of user "reputation" support this view.
We're betting on the structured input approach - i.e. suppress the noise from the start. We wish Twitter good luck, but bet that in the end they'll be forced to expand hashtags to more structured categories, and enforce some level of a priori submitter "certification".
Nick DiGiacomo
Co-founder, Vanno
http://www.vanno.com/
Even the best system I can imagine (a second-by-second ranking of each tweet based on each person's independent assessment of the value of the tweet) would work against Twitter's central idea: a foundation of egalitarianism, the efforts of people to stand out from the noise, the efforts of people to pan-handle the noise for good information, and an element of luck.
Reputation with respect to some specific tweet is very much a contextual issue and what may have warranted high ranking even a moment ago may no longer warrant it now for a whole slew of reasons. I do no look forward to the day when my stream is overwhelmed by highly ranked tweets all saying the same thing because those "in the know" are only talking to each other. I could go to the dailies for that.
I think the best method for this assessment of the value of a tweet is the ever alert folks who are doing that right now and pan-handling the stream quite effectively. For example, if I know someone who should know something on some issue I seek out their tweets -- I don't need an algorithm to decide for me. A better search tool would help, but based on content not reputation.
This kind of search will turn Twitter quickly into the preferred search-engine for all "fresh" information – and maybe even make bookmark services redundant in the long run...