Is Twitter A Substitute For Set Top Box Data?

For the past decade, the cable and television industry has been working to use set top box data in the way that click stream data is currently used on the web. To some, it is seen as the holy grail and/or the savior of the television business. The thinking is that if TV viewers can be understood, segmented, and targeted in the way that Internet users are, the TV experience and the TV advertising business can be transformed in ways that are beneficial to everyone.

The problem with set top box data is that it is controlled by the cable companies and they have not been particularly open with it. There are a few companies that have access to it, but suffice it to say that building a business based on getting access to set top data is not for the faint of heart.

So I was quite interested when I came across this post from one of our portfolio companies, Simulmedia, which is working in this TV targeting space. Simulmedia was founded by Dave Morgan, founder of Real Media and TACODA and a veteran of the online ad targeting business, to deliver similar products and services to the TV market. So he knows a bit about all of this stuff.

The post describes a little experiment Simulmedia did to analyze TV channel surfers using only Twitter posts as their data source. Simulmedia explains why they turned to Twitter data for this work:

To investigate surfing behavior further we tapped into data from Twitter,
where dozens of people announce to the world that they are channel
surfing every day.  Unlike traditional ratings or second-by-second
set-top-box information, Twitter has rich qualitative information that
reveals motivations behind viewers’ surfing.

Here's the results of their study:

While the data in and of itself is interesting, I am more intrigued by the idea that Twitter can be used as a panel that once it is large enough (and maybe it is already) that can produce much of the same data that is currently locked in set top boxes and controlled by the cable companies.

The Internet is disruptive and reminds me of that fact every day.

Reblog this post [with Zemanta]
#VC & Technology

Comments (Archived):

  1. Pierre

    Fred, have a look at when a popular US or UK tv show is on, and you’ll realize that more than 50% of the buzz on twitter is about these shows through our real-time tag cloud. You could even predict the winner of Britain’s Got Talent just by looking at the tag-cloud!The same applies to popular conferences (Apple Conf, E3, Google I/O) when you can study feedback from the crowd in real-time. You can see what products / annoucements grab users attention, and we find it very powerful.

    1. fredwilson

      Thanks. I’ll check it out

      1. Pierre

        If you get the chance, give it a try right now. The Apprentice is on air, and interesting things should start to happen to our tag cloud.

  2. Matt Millar

    The challenge with Twitter is both in the effort required for a viewer to type a sentence, and also being able to analyse that data quickly enough. While Twitter does give a great timeline of intentions in real time, it’s costly to analyse.Imagine if you could build a system that was even easier than twitter – and have views aggregated in real time – what impact would that have on understanding audience behaviours? It’s a fascinating space, which is why I’m building a solution to address just that – let’s see if I can use the internet to disrupt live TV broadcast and live events.

  3. Siminoff

    So cool… I wonder when/if the fortune size companies will realize that they can create better survey results using platforms like Twitter then trying to directly bother their customers for the data.

    1. fredwilson

      I wonder if this will impact the survey business

      1. Siminoff

        Not sure if you knew this, but I really don’t like the survey business:)

        1. fredwilson

          Not surprised

  4. ErikSchwartz

    What’s interesting about implicit data is that it is implicit.Tweeting is very explicit. Even more so if people think their tweets can influence others.

    1. fredwilson

      Is that good erik? Or is it a mixed blessing

      1. ErikSchwartz

        I bet the audience for “the real housewives” under counts in self reporting panels, people want to look smart. Look at how PPM data compares to arbitron data. I also think some formats of television tend to lend themselves more to tweeting than others (reality TV over dramas). Now add into the mix that the producers will know that encouraging viewers to tweet can raise their ad rates… You’ve got a real mess.I understand the power of measuring “influencers”, but I think there’s too much chaos here to replace STB data. I think this can supplement and add some context to STB data, but not replace it.

        1. Graeme Hein

          People lie in self reporting, especially something that is as public as Twitter. It being public drastically changes people’s actions, as you can see by looking at #roft and #pcpo.You can do interesting things with Twitter, but analysis that needs objective data that is replicable across types and elements of programming can’t use it effectively. Too much tweaking required, where arbitrary adjustment parameters would overwhelm the signal.

      2. ErikSchwartz

        Apropos;…”Conversely, Sequent’s Spaeth said traditional TV viewing has been “pretty drastically under-reported” by research that asks people how they consume video. The reason why, he said, is that research based on how people perceive they consume media isn’t nearly as accurate as research that actually observes how they use it.The ad industry historically has known about such “halo effects” and the fact that it is considered socially unpopular for people to report that they watch as much TV as they actually do. On the other hand, Spaeth said people tend to over-report their online and mobile video consumption, because “it is new and cool.” “

        1. fredwilson

          I wonder if biases like that are actually signals of where things are headed

          1. ErikSchwartz

            I think the biases are hugely important. But to leverage bias you need an unbiased control sample.Biases are about who we want to be. Biases are important because marketing is more about who we want to be than who we actually are.

  5. Louis Adekoya

    Just over 2 years ago I suggested at that Twitter could be used to gather TV audience data Glad to see it happening

    1. fredwilson

      Man you were prescient. Got any other predictions for us?

      1. Louis Adekoya

        I can’t decide if you are being facetious or not here Fred πŸ™‚ Nevertheless, I’ll play along. I have posted lots of tech ideas to my blog but I think the closest I have come to making predictions is a post titled “Tech Segments Requiring Innovation in 2008” – I’ll let you judge how well or badly I did …I love the word prescient by the way – always have for some reason.

        1. fredwilson

          I was being totally serious. I’ll check out the link. Thanks

        2. Mark Essel

          Fred’s great that way, he’s almost always looking for new ideas or more information to improve his views. I think he has a knack for creating value from any genuine information. As an added bonus regular readers get new vocabulary words from time to time πŸ˜‰

          1. Louis Adekoya

            Best of all I think is him really joining in the conversation the way he does. And I say this under no duress nor in exchange for any promise of venture funding πŸ™‚

  6. Dave Morgan

    Matt … very good points, but you don’t need a system “even easier” than Twitter to change the television industry. 140 characters of intention/action/mood data per tweet is much more robust than what a set-top box can tell you on a daily basis, let alone per event. The real power and opportunity to change the television industry comes from building predictive models that leverage a number of disparate data sets, Twitter, set-top boxes, program guides, weather, news, events, etc. Once they are put to use – to enhance and improve the consumer television experience – we will see a lot of changes.

    1. Alan Warms

      Dave, you can also layer in data from viewing habits online that could also help you. I just got involved with a company I think could be interesting here — shoot me a note

    2. Matt Millar

      Dave – agreed – I think there are two routes for innovation here:1. Building data models that leverage existing data sets for new areas2. Building new interaction models that drive additional dataThe two interact – in your example Twitter have built a new interaction model that you can leverage – but there are other innovations & and more leverage from building new interaction models too.There is a cycle of innovation here – more interaction generates more data which informs more innovation.The opportunities to integrate this sort of information into IP connected STBs is fascinating, as this gives another point of “internet reach” for feedback and information in the living room.

    3. fredwilson

      Yo dave. You are not supposed to be laying out your business plan in a blog comment πŸ™‚

      1. Dave Morgan

        Fred, you and Brad have shown me the light of open source business planning … can’t go back now πŸ™‚

        1. fredwilson

          Indeed. This comment thread is a proof point of the value. Free consulting FTW!

          1. Mark Essel

            You guys are cementing the value added from open social design. Just figured you should know Fred, you instigated an open access/source trend post (didn’t see a track back but there’s the link for all who read AVC and dig the OPEN vibe. http://www.victusspiritus.c

      2. jarid

        Right, he’s supposed to do it via a tweet. :)Hmm, business plans in 140 characters or less. Someone is probably already working on that idea…

        1. fredwilson

          hey jarid. nice to hear from you. check this out about our baby freemium…

          1. jarid

            Yeah, she’s all grown up now. Seeing the recent articles referencing freemium in the WSJ and NY Times made me all faklempt. :)I just wish I trademarked it like Riley’s three-peat.

          2. fredwilson

            I just wish we had a basketball team in NYC

    4. hypermark

      I think that you hit the nail on the head. While some would argue that there is very little data/context within Twitter, services like StockTwits, in conjunction with client apps like TweetDeck show that with a very simple handle tag ($ in case of stock symbol in StockTwits) you can not only aggregate and cultivate a meaningful, focused conversation that spits out on the web, but provide customized views of same within a client app like TweetDeck.Specific to the TV experience, that suggests value added services for cultivating engagement around live programming (sport events, reality TV) and popular shows (Lost, Desperate Housewives), programming recommendations (Watched/Related/Popular/Favorited/NOW), and of course, better targeting and information overlay.As a total sidebar, I blogged about this some time back in a post called:Middleband Channels: Television content-aligned services irony is that the post is dated 2005, which only goes to affirm the fact that there is a certain inevitability to many ideas such that the hard part lies in actually making them work. :-)The power specific to leverage of Twitter is that it is both out of band (i.e., unbounded from a special purpose box) and an open platform (unbounded from a specific TYPE of service), which of course enables disruption of a lot of vested interests.

  7. Ed

    I fear that as market readers, we are too fixated on one metric; the market. Much wisdom can be gained for a deep look into the individual, passionate user.(And they deserve to be understood. They are not the crowd. They are simply in it. Driving.)

    1. fredwilson

      Can we ever really get to one to one when it comes to mass media?

      1. Ed

        Not with each and every user, no. But a deep understanding of some passionate usersis enlightening. In fact we can grow and accelerate our discernment this way.The CEO with his ear to the ground during a stampede hears only thunderous noise.However-the one who accepts dm’s from [listens to] the passionate “nobody” understands himself and his product better than he ever could alone, [or through only staff and “consultants”]. I know of the existence of such relationships.I’ve seen the early fruits. It’s worth budgeting time and money for when you find a giving tree.

    2. Louis Adekoya

      Ed, your comment brings to mind an interesting HBR article I read recently on the power of ‘psychographics’ over demographics – learning more about what influences your potential customers’ buying decisions etc as opposed to just what market segments they fall into.This said though, I think Twitter, Facebook etc do absolutely help with this.

      1. Ed

        Precisely. I suppose Facebook also, but Twitter is a particularly tremendous listening tool.Let’s assume for a minute we’re talking about the CEO [or ANY employee] of a singleentity [although it is equally effective for a VC like Fred with several interests to track,but inherently more time consuming].This CEO will hear an instream of sophisticated data, packaged in prettypresentations from his/her immediate staff and vendors.It is irrelevant whether we’re considering Tony Hsieh, Richard Branson, or Evan Williams,Twitter CEO himself. This applies to ANY corporation large enough to be discussed on line.[Because soon, all will be on Twitter, and local will be as relevant as global]In the course of reading reports from your VP, or Forrester, or accessing in-houseanalytics directly, there is an opportunity, and no replacement for, one on one with your customers.We didn’t have this before Twitter. Yes, there were blog comments, forums, live events, etc.But now, even the isolated billionaire can plug in directly to the sentiment of thosewhom she wants to understand.It seems so obvious. But most are just not good at it. Envision a Richard Branson on his laptop on the couch, using to monitor the discussion about Virgin’s next launch.He reads many comments over a few days. Some surprising, and/or confirming themesappear. Perspectives and perceptions which he and his illustrious team never considered are a window into the consumer, peers, competition, and themselves.But- He takes note of the most thoughtful. Whether passionate, deeply analytical,personal, or simply original; if he has insight, he’s got a million-dollar compass.If they get hung up on “I don’t know this name. No formal analyst”, they missed the whole point.I’m fortunate enough to have several successful business people follow me on twitter. I could offer advice to these accomplished men and women because twitter exists. Here’s the key: some have rebuked my observations [remember this isn’t about me],because they’re not used to listening. Many come back later with “you were right”, oncethe masses displayed the direction forewarned. Others don’t admit it, but you can observeimplementation of the ideas, or further engagement in the discussion.I do not mean to hijack Fred’s post, but I have put hours of thought and observation into this.Even the excellent, aware ‘new school’ managers who engage, could glean far more if they knew how to listen to one or some deeply, by following up and asking, rather than only skimming the top OF 1,000. [Thank you for space on your blog Fred. I appreciate the opportunity for conversation].

        1. fredwilson

          I love the image of branson searching twitter for mentions of virgin. He would be pleased if he did that

          1. Ed

            I Tweeted @RichardBranson 3 times, and replied on his blog 3 times. If he had listened to me once, he would have had a stunningopportunity. Every major player I’ve asked has responded”Most brilliant move I’ve heard in years”. But he did not. He keeps paying millions for mediocre instead.

  8. Mark Essel

    The influence of information availability and it’s clever application to existing business challenges is an awesome ongoing transformation we are fortunate enough to witness firsthand. What wonders will our next generation take for granted? What bigger problems can we continue to chip away at with the free, open and relevant information? Thanks for the post Fred, always worth the time to share in your discoveries.

  9. $3236

    In the late 90s I had a business plan set to go where the click stream data was collected, stored and transmitted by a universal remote control that the company would give away. This was prohibitively expensive 10-12 years ago but it would be doable now. The universal remote eliminates the set-top box data-lock-up problem and wireless transmission from the device (wifi or cellular) sends the data pretty much in real-time. You would know the demographics of each household that received one of your free devices and, hopefully solve part of the targeting problem using the sample of households who had the device.

    1. fredwilson

      You could do that with an iphone app

  10. mediaeater

    We have been using Twitter this way for a while – Looking at prime time hours and the number of posts leading up to and during a show, comparing them against others in the same time slot or other like shows has been very interesting. We plan on dropping Trendrr out of beta in two weeks one of the features allows you to render TV data this way. What is cool is when you start to look at Twitter posts this way it becomes predictive in some aspects and an essential part of the visibility needed for a network, producer or a show.A late night snap shot for example…You can also use Twitter to trend the popularity of a show over time… is some gold in ‘dem there tweetsAs an analyst this is great fun, and valuable insights, as a marketer it is really neat to see how you can enable the conversations further at crucial points affecting the awareness curve and on-air tune-in.

    1. fredwilson

      Trendrr is getting more and more interesting as more open data sets like twitter emerge

  11. andrewparker

    I’m often reminded of TV worth watching by Twitter. While I’m checking up on Tweets I’ll see something TV-related trending, such as the Bull-Celtics matchup earlier this year, and it will spark me to go watch it. Or, someone from within my network will be live (or TV-live) tweeting an event that’s on TV, and it will encourage me to go see what he’s looking at.So, Twitter is a data source for both segmenting users (as Simulmedia did), but it’s also a data source for discovery.Can’t wait to see these new computer-embedded TV (or TV platforms) display a Twitter feed next to what I’m watching, filtered by the content I’m watching.

    1. fredwilson

      twitter’s bd person for media is working very hard on this stuff

  12. Steven Kane

    love this concept, generally-speaking but have to differ on the specifics and practical sidenobody ever wanted or tried to use set top box data to target “the way that click stream data is currently used on the web”set top boxes *at best* can be used to target *households* – not individual viewers(clickstream data can rarely if ever be used to target invididuals but that’s another matter)further, set top boxes explicitly, and in detail, and with user consent, allow for highly granular targeting (again, of households). thats because every cable account is a paying account, and so every set top box knows where it resides, and that information all by itself tells the vast vast majority of marketers nearly all they need to knowzip code (from which everything else can be derived)affluence/income levelauto preferenceslikelihood of petslikelihood of childrenlikelihood frequency and type of traveleducation levelsetc………the real rough and tumble set top box data business has been audience measurement — exactly how many households are actually tuned in to any network or show or ad?and not surprisingly…1) the data can be easily and highly productively harvested parsed and mapped for that purpose2) none of the powers that be (networks, producers, advertisers, ad agencies) really want to know so the business never gets off the ground (because the knowledge would only destroy the existing food chain, not enhance anyone’s position in it)totally discreet individualized ad targeting is a holy grail – worth pursuing, probably never achievable, and probably not worth it even if achievable. successful marketing is as much about making converts as preaching to the choir!

    1. fredwilson

      I hope dave morgan sees this and replies. You bring up a lot of good points that he’s been dealing with

    2. Patrick

      Agreed, but I am not sure discreet individualized perfectly customized exact targeting is truly worth purusing, just give the end user control over the relevancy of the ads and I think we are getting to the right spot.

    3. Dave Morgan

      Steven, great points, but TV doesn’t need “perfect” targeting to be made much better. As you point out, as much as ad agencies and marketers like to talk about how much they would like to target their ads with sophisticated socio-demographic target, very few of them have the capacity or will to follow through. We believe that many user-derived data sources – whether it is from set-top boxes or from Twitter or from weather reports – can be used to better inform the delivery of marketing messages to consumers. True, one of the first uses of set-top box data will be for better audience measurement. But you can also use it for better ad targeting, and you don’t need to wait for household level addressability to do it. There is a lot of value in mass customization and a lot of value in creating those customizations with user-derived data. Today, there is virtually none of that in the $70 billion spent in the US on TV ads – quite a lot of room for improvement.

      1. Steven Kane

        “Mass customization” sounds cool – can you explain a little?

        1. hypermark

          Mass customization is a model whereby you can use a systems-based approach to customize a product or service to meet the specific needs of an individual as opposed to generically serving a broad demographic or a market segment.Think CafePress, where you can build a custom t-shirt, pick, size, position logos and add custom text; or a credit card company that can send out 75,000 different highly targeted offers as low cost “seeds,” select and amplify the winners (based on consumer response rates) and toss out the losers. Instead of one size fits all, it’s one size fits ME.Here is a short post that frames the marriage of digitization and systems thinking into mass customization offerings: it out if interested.Mark

        2. Dave Morgan

          In the context of what we are doing to optimize television program promotion, “mass customization” means segmenting TV viewers into groups of millions of tens of millions according to their viewing preferences and correlating that to the programs or networks or time slots that they are loyal to or the geography where they live, and then delivering an “on-air” promo to that entire audience, not just to a single set-top box (which is not practical at this point in time). While you won’t have delivered a promotion with perfect, personalized relevance, you will certainly have delivered one which is much more likely to be relevant – and thus drive the intended response of causing the person to watch the promoted show – than promoting the same shows to the same people on the same programs irrespective of their interests or where they live.

      2. fredwilson

        Nice. I got my wish!

        1. harpos_blues

          I’m late this conversation, but I”m wondering how and why the the proposed “mass customization” differs greatly from the data collected by companies such Rentrak — the market leader in this space. As an example, Rentrak provides real-time box-office numbers for the film industry, among many other data sets.I met couple of executives from Rentrak a while back, and was impressed with the vast amount of data collected. The sheer volume of transactions they process would make a credit card company blush.However, to my knowledge, Rentrack does not provide an advertising delivery system. We discussed API’s during that conversation, so in theory there may be a mechanism to access Rentrak data, I doubt it would be “free” (as in beer), though.

          1. fredwilson

            I’ll forward this to dave. He should meet them if he doesn’t know them

  13. Rob K

    Interesting. I have always wondered why the big cable companies didn’t put Neilsen ratings out of business. Why bother with a tiny sample when you have the actual universe behavior?

    1. ErikSchwartz

      Because advertisers want 3rd party measurement.

  14. maxkalehoff

    It’s been almost four years since we built “TV BuzzMetrics” (my last startup company) which Nielsen subsequently bought (…. Online expression already delivers incredible value in delivering vast qualitative insights, but there’s a LONG way to go for it to influence or become part of the currency of buying and selling audiences.

  15. csertoglu

    Only to the extent that users tweet, and stick around. The recent HBR and Nielsen data makes me question this.

    1. fredwilson

      I’ve been quiet about all this nobody tweets and twitter quitter stuff but since you raised it, let me tell you what I thinkAll blogging systems, and twitter is very much a blogging system, have ratios of contributors and readers that average an order of magnitude difference. A lot more people are reading tweets than sending tweets, particularly the celeb following crowd that has come on twitter recentlyAs for twitter quitter, when we made our investment only about ten pcnt of singups stayed. Now its forty. And I predict it will be sixty or more in a year. The “out of the box experience” was awful a year ago, its better now, but still needs improvement and its getting a lot of work that nobody has seen yetEv talked about this in the D interview and did so very articulately

      1. Jeff Shultz

        Doesn’t the fact that Twitter data creates (at best) a representation of the actions of a subset of consumers (contributors), and not the larger Twitter universe (readers) or the actual TV viewing audience (Twitter users and non-users), bias the data? I would think the bias would be significant enough to invalidate the data as a means for divining the actual habits of the general viewing population.Validating the sample is the core value of any audience measurement service. STB data is attractive (among other reasons) because it can take the form of actual census data w/o the possible bias of an extrapolated sample.

        1. Nick Molnar

          Some of this could be fixed by having Twitter profiles be more in depth/structured so that you could normalize the tweet activity to the population. Right now, Twitter TV trends primarily tell you what affluent, late 20s to early 30s, professional males are watching on TV. That’s valuable stuff, but it’s not a replacement for STB data.Separating the other demographics that are tweeting is just too hard with the current profile setup, so the data is biased towards the mean (the young, affluent, males).Facebook Status Updates on the other hand could give demographically adjusted results all day long because of how much information (age, sex, location, religion, employment, taste, etc) they have on their users.

          1. fredwilson

            Agreed and twitter has this on its roadmap

        2. fredwilson

          Today for sure. One day, maybe not

  16. courtney benson

    Now all we need is some great middleware to bring it all together!

  17. J.D. Falk

    Cool! Only problem is, if I tweeted about every lame show that I decided not to watch, it’d get really annoying and all my friends would stop following me.A video equivalent of might make a lot more sense, particularly if time-and-location-shifted recordings could be included.

    1. Patrick

      Interesting. But I think far too often we are looking for a complex solution to a simpler problem of content relationship. Rather than worrying about set top data or how I am twittering, allow me to select a myriad of categories that are relevant to me so I can receive my content as ad supported by marketing that matters to me. I can opt in and opt out of advertisers based on their ability to stay relevant. I can set times where specific advertisers can place ads, ie I am watching a baseball game with my 8 year old, I really do not need to sit and watch the dieing smoker commercial anymore or penis enlargement thanks very much.

      1. fredwilson

        We all wish for that but I wonder how many of us would take the time to actually set up such a profile. I think we can there by allowing marketers to have access to implicit data about our interests and activities. That would also have the benefit of being a real time and dynamic profile

    2. fredwilson

      You can watch through boxee and get that today

      1. Patrick

        I know, but you have to be a Mac (which is fine for me at home) or Linux user at the moment, and Cable providers have the pipe, content and scale to make this work for both sides of the house without reliance on a os platform. As a user of course it is about me, as the marketer me it is about real scale.

  18. kidmercury

    i think twitter is an aggregator media company (i.e. it collects everyone’s micromessages), and the next wave of innovation will be in companies that can filter/remix aggregators based on niches. so a set top box data company, meaning a company that understands how to use/interpret set top box data and deliver the right data to the appropriate members of the community, is where value will be. but such companies will use more than twitter, they will use all aggregator companies (youtube, delicious, amazon, technorati, etc). i wonder how much oppty there is for aggregator companies here, seems to me they will be a cost center for the remix companies, and that the real disruption/value creation oppty is in the remix companies.

  19. lucaf

    Interesting thought, but the challenge of media measurement is that your sample has to be demographically and behaviorally representative of the universe. Twitter users are not a good sample. And I’d argue that twittering about media consumption is indicative of low engagement, hence your measurement misses the most valuable eyeballs.

    1. Patrick

      I don’t think this is meant to say Twitter is the “now” for media measurement, but honestly can you say with any degree of confidence that Nielsen measures TV both demographically and behaviorally representative of the viewing universe? The diary method might have been great in 1975, but it really fails three decades plus later.

  20. RacerRick

    It would be cool to have an application that tells exactly how many people are watching what TV show at a set time, based on Twitter posts.

    1. markslater

      why? this is the closest thing i have heard of to a wives knitting club in years…

  21. markslater

    Just so that i am clear, you sit in front of the telly and tweet what you are watching? really?this is why the mainstream has a hard time ‘understanding’ twitter.

    1. Patrick

      People stand around a water cooler or coffee machine and talk about what they watched on television. Really, they do. Is this so different? Does it make it the single definition by which to understand twitter, hardly. And I don’t think defining it singularly for one purpose will ever work.

      1. markslater

        this speaks to the very core – you stand around a watercooler because you are physically having a chat with someone. This involves gestures, tones, movements, actions – all combined to create a conversation the result of which has some type of derived meaning for each constituent part. Juxtapose this with tweeting and i would argue they are radically different.

        1. Patrick

          Of course they are radically different one is after the fact with accrued information and reason to converse, the other is a statement in the moment, and frankly more than one television show is going to drag twitter, facebook etal into their viewing pattern for the audience next year. The news channels here in the US do it, radio stations are doing it, print publications do it, bloggers do it, networks/producers are going to try and make it happen for original programming too. Hell they will do it in syndication as well.

        2. fredwilson

          Yup. Twitter is way better because the conversation scales better. Same with blog comments

          1. markslater

            so the value is the scale of the conversation?

          2. fredwilson

            To me it is. Why settle for a discussion with a few ppl when you can talk to thousands?

          3. markslater

            depends on what you want from a discussion Fred.universities are gonna be studying the psychology of tools like twitter.

      2. GraemeHein

        Fark has had 24 threads that people would read and post while the show was on and afterwards. Commenting about minute to minute developments, posting images (including captures) in response to plot developments…The threads were VERY active 2 years ago, so people really will do this. Now of course the FARK community is rather technically engaged, but not that much.It’s really a generational thing, especially if everyone in the house has a personal laptop. My own engagement with media changed substantially after I switched from a desktop to a laptop (well multiple laptops and a few desktops but..). It also depends on being alone or with housemates. I do not use my laptop like that when my girlfriend is on the couch with me. Not even a demographic issue but who is around while you’re watching and what else you’re doing.

    2. kidmercury

      lol, i dont get it either, wtf is up with these peopleIMHO rapid micromessaging/chatting needs the right context. twitter has no context, which is why there is a lot of “wtf is this” amongst non-geeks/twitter newbies. i wonder if a niche community built around a tv show that had a twitter-like app as a part of it would fare. i think that might get hardcore fans live tweeting about it more.

      1. fredwilson

        Yup. Look what stocktwits has done for traders. This is the future of twitter. Its a platform not a community or an app

        1. markslater

          i agree Kid – i think that this is what i struggle with.i have seen at least 3 services recently that are taking a kind of anti tweet approach – where the focus is on highly concentrated groups with alot of context. Kinda like micro-social-network applications.Has twitter shown the masses the means by which they can re-constitute their fraternal instincts again – and what does that mean to twitter?

          1. fredwilson

            Twitter won’t but the apps and services built on top of the platform will. Twitter is like smtp or sms. Its a platform or a protocol. Its getting built into mobile OSes that way

    3. fredwilson

      I do that at times but not frequently. Hundreds of thousands do though

  22. Geoff

    Fred -w hilst trying to reach this page in an internet cafe in Peru your blog defaulted to this URL… the message was;-Not FoundThe requested URL /ig/modules/translatemypage.xml&up_source_language=en&w=160&h=60&title=&border=&source= was not found on this server. Their computer was on XP with IE6A bit annoying πŸ™ good job I bought my Eee with me

    1. fredwilson


  23. joeagliozzo

    Not to hijack the thread, but I recently starting using the new Hulu Mac app and it is fantastic. Not quite a tv (hd) like experience- but through a Verizon FIOS pipe and on a 42″ flat screen, pretty darn good. This does seem like the future if the quality can be bumped up even a bit further. Then insert targeted ads and you probably have a more profitable (but obviously not yet larger) system for advertisers. End up with the PPC of TV industry, along with all the benefits, auction model based on ROI, tracking, targeting, etc.

    1. fredwilson

      Yup. Boxee. Hulu. Front Row, etc. That’s where tv is headed like it or not

  24. rod edwards

    Fred, Dave – how do you account for adverse selection in using Twitter derived data in a decision-making setting? it strikes me that that is the larger question.Set top boxes are passive, Twitter participation is active – that is to say, Twitter metrics will select for certain user classes. Identifying those classes and extrapolating them into larger population trends (i.e.: blending twitter metrics with traditional metrics via some credibility system) will be the challenge that takes something like this from “curiosity” to “core” and “actionable.”I see a lot of parallels to the statistics and mathematics applied in the insurance industry (experience analysis). If you’re working along these lines, you may wish to consider retaining a consulting actuary.

    1. Dave Morgan

      Rod, you bring up a good point. Not all data is the same … some is observed, some is reported, some records events, some records intentions, some represents beliefs, some represents unavoidable externalities. Aggregating this for specific vertical uses, along the lines suggested by kidmercury, requires dealing with the different data differently. It requires a vertical, ‘point of view’-focused combination of mathematical modeling, discipline, creativity and luck. I like the actuary idea … our team includes a behavioral biologist, modeler of credit-risk derivatives, a brain scan researcher and a domain specialist on entertainment programming.

      1. rod edwards

        Dave, thanks for sharing your thoughts. I don’t know enough about the skillsets of each of the team members you noted — the modeler may very well have things covered in regards to the statistical rigor of source-blending.Whatever the case – best of luck!-R

  25. Michael Johnston

    The problem with this method of data collection is that the results are subject to poisoning by those who would seek advantage through affecting the outcome.

    1. fredwilson

      Ie spammers. Excellent point and something twitter and all social systems struggle with daily

      1. rod edwards

        Not to overhype the connection to the finance industry, but I wonder what a company like Angoss would have to say about identifying spam?'s essentially what they do for financial risk; identifying claim spam for insurance companies, etc.

  26. Morgan Warstler

    A number of things here -First, Twitter isn’t going to be competing with set top box data, but certainly with current Nielsen data. As Dave is certainly aware, the intense stream analysis coming from a Visible World is doing more than just helping to program commercial pods (the blocks of ads in between the other stuff on TV). If it is knowable the cable companies will soon know it. Cablevision and Comcast are both testing personalization “to the household” as we speak. And knowing who in the house is watching isn’t as hard as you might think. Frankly, with TV, it isn’t about lack of technology – it is about lack of agreement between the stakeholders (content, advertisers, cable). Witness the ongoing discussions with Project Canoe (big fan here).Also, using as a leading market example – there are about 62 lifestyle clusters (claritas data) – really solid personality types spread out over the nation’s zip codes. And I know it feels like there is an infinite amount of video programming – but there really isn’t. So finding certain lifestyle type like “Bohemian Mix” or “Young Digerati” or “Middleburg Managers” out there in the TVerse isn’t actually as hard as you think – they’re all each watching only a couple of shows at any one time.This brings me to the DVR, which with companies like INVIDI makes much of this future targeting stuff just so yesterday. Soon enough, the commercial is going to be on your DVR Fred, and it wont matter what channel or stream you are watching, you will be seeing the commercial exactly the number of times the buyer paid for -before it is deleted.Last note: Boxee should start to experiment with “telescoping ads.” Gain some early expertise.Last, last note: Big fan of Simulmedia. Our company has a different approach to mass customization.

    1. fredwilson

      Thanks for this comment morgan. Lots of interesting stuff to chew on

  27. bombtune

    How do you think Internet disruption will threaten China’s protective Internet infrastructure. Here’s my perspective on the music side.

    1. fredwilson

      Really good question. As a big fan of freedom, I’d like to know the answer. I’ll read your post

  28. howard lindzon

    i bought tivo stock today…does that mean anything ?

    1. David Cooperstein

      This is a great conversation – there are so many ways that a twitter-like stream could help inform and drive audience research and data. Two thoughts on this:1) If any of you have played the new Spymaster game on Twitter ( it is an example of getting passed active use of twitter and into more game-playing/group think/passive contribution while using the community. Imagine if you could embed this concept into the set top box, so that you can get twitter-like commentary and activity based on the show being watched (or being switched off) without having your computer/Eee/iPhone etc. on your lap. Watching reality shows with voting and community participation would be more scaleable, and the results mapped back to the audience. Twitter may do this at some point, but the STBs need to get to the next level as well.2) Where is the Nielsen angle on this? They are getting harangued by the industry for having trouble measuring viewership. But if they could tie the twitterers to their audience measurement panel, they could start to normalize the data.The one thing about TV (like the paper) is that there needs to continue to be some exploration, which is what I believe Simulmedia will help. But the research will need to come from something no less complex than a Quantcant multi-variate audience analysis with factors for active and passive actions.

  29. William Mougayar

    Aside from the TV-Twitter relationships that have been well articulated here, Twitter as a polling mechanism holds promise especially for large companies and their customers where users are well identified and profiled. Remember when Facebook stole some thunder at Davos by showing 120,000 responses in 20 minutes, live on the stage? I just found this interesting Twitter Polling service:

  30. pcervieri

    here’s a video of dave talking about the article and john gibs from nielsen commenting as well at the Advertising Research Foundation conference:

    1. fredwilson

      Thanks for bringing this video to my attention. I’ll watch it