What’s Your Google Viral Score?

Will Google Plus show advertisements one day? If they do, how will they decide upon the ads to show different users of the social network? A 2010 paper, AdHeat: An Influence-based Diffusion Model for Propagating Hints to Match Ads (PDF), described one method of advertising on a social network that was actually tested on Google’s world wide (except for the US) set of Q&A type sites with the code name of Confucius. It also incorporated the Confucius User Rank into displaying those ads. The user rank approach to reputation scoring for Confucius, and for choosing advertisements for users of the system appears to be an method that would work well in deciding upon reputation scores for users at Google Plus.

A Google patent granted in early December, 2012, provides a different approach for showing advertisements and other content items to users of a social network like Google Plus. The patent makes it clear that while the approach described within it might be used for advertisements, it might also be used to show other content as well.

Patrick Henry at the Virginia Assembly, giving a speech that spread quickly across the colonies.

The content displayed to a user would be based upon a viral score, or “content propagation likelihood,” for users, which is the likelihood that content they share, (broadcast, reshare, comment on, endorse, like, recommend) might be spread throughout the user’s social network. Content in the patent is referred to as an entity (e.g., an item of content such as a video, audio clip, news article, photograph, etc.). In most patent filings and other writings I’ve seen from Google about entities, the authors have been referring to specific people, places or things. So, it’s an interesting twist to see documents and other content types on the Web referred to as “entities” as well. Considering they are specific things though, with specific addresses (or URLS), and have their own sets of attributes and properties associated with them, I don’t find it odd to refer to them as entities.

I’m still trying to ferret out exactly how this patent might be applied to content other than ads, but the patent does tell us:

While the following description includes many examples presented in the context of advertisements and makes reference to a user’s “ad propagation likelihood,” the scope of the present disclosure is not limited to advertisements.

Instead, embodiments and features of the present disclosure are applicable to various content items in addition to or instead of advertisements, and thus reference may also be made to a user’s “content propagation likelihood” as well as similar terms or phrases.

The patent is:

Method and system for selecting content based on a user’s viral score
Invented by Ping Wu and Jennifer W. Lin
Assigned to Google
US Patent 8,332,512
Granted December 11, 2012
Filed: September 27, 2011

Abstract

Methods and systems for selecting and presenting a content item, such as an advertisement, to a user of a social network are provided, where the content item is selected based on a calculated “content propagation likelihood” for the user.

A user’s “content propagation likelihood” is a likelihood that an entity (e.g., video, audio clip, photograph, etc.) will spread throughout the user’s social network, and the social networks of the user’s friends, when the entity is shared (e.g., broadcast) by the user.

A user’s content propagation likelihood is computed using weighted measures of various ways in which an entity can spread through a social network. A user’s content propagation likelihood may also be set for a given vertical (e.g., music, sports, etc.) and/or a given media type (e.g., images, videos, etc.) that pertains to the particular user.

Share and Reshare Scores

A user’s ad propagation likelihood may be calculated using a “share score” and a “reshare score” for the user.

Share score is the probability that an entity (content item) will spread throughout a user’s social network, and the social networks of the user’s friends, when the entity is “shared” (e.g., broadcast) by the user.

Reshare score is the probability an entity (content item) will likewise spread when the entity is “reshared” by either the user, a friend of the user, or another user in the user’s social network.

In a “reshare” situation, the person in question may be someone resharing information over a social network instead of being the one who originally created it or brought it to the social network.

Calculating a user’s “share score” could also involve how connected someone might be in a social network, such as looking at how many other people that person might be connected to, or have some kind of relationship with.

Vertical Types

A content propagation likelihood could also be set for specific verticals, such as music, technology gadgets, sports, etc., and/or for a given media type such as images, videos, audio clips, that pertain to a particular user.

Someone who frequently shares content related to music in a social network, and who has friends in the social network who often reshare such content initially shared by the user, may have a high content propagation likelihood for music and other related verticals. This high likelihood could be used to decide that user should be shown advertisements relating to music.

Someone who likes to share photographs might be shown image-based ads. Another person who shares a lot of videos might be served with ads (or other content items) in video format.

Entity Usage and Advertisers

Entity usage information about different content items might include information about:

  • Whether or not a user has indicated he or she endorses (e.g., likes, recommends, suggests, etc.) the entity
  • Has posted a comment about the entity
  • Has sent a message (e.g., a text message, e-mail message, etc.) about the entity
  • Has visited a web page associated with the entity
  • Other information related to user-behavior regarding the entity.

If the content types in question are advertisements, an advertiser might limit the display of the ad based upon certain criteria as well. These might include whether or not the user is:

  • Located in a certain geographic region or
  • Only if the time of selection and presentation falls within a certain date range (e.g., between a start date and an end date indicated by the advertiser).
  • Above a certain age (e.g., a child under the age of eighteen or a user who has not yet reached age twenty-one).

Personal Information Collection Optional

Under the patent, the collection of personal information about a user would be optional, and they might be given the option to not have his or her personal information collected or used in any way.

So a user would be given the change to opt out of information such as their geographic location, preferences, and similar information, or it might be hidden by removing personally- identifible information, or generalizing geographic information so that the location of a user can’t be determined.

Entity Usage Information Collection

A screenshot from the patent showing how entity usage information might be collected.

An entity, or content item, might have information collected about its usage, that can include the number of:

  • Endorsements (e.g., “Likes,” “Recommendations,” “Annotations,” etc.) it receives by users in the social network,
  • Reshares of the entity by users,
  • Comments made about the entity,
  • Messages sent mentioning the entity,
  • Visits to a web page or physical location (e.g., store) associated with the entity,
  • Purchases of a product or service associated with the entity, and/or
  • Search Results Clicks on the entity.

Across Different Social Networks?

The patent tells us that entity (content item) usage information collected isn’t limited to actions and behaviors of just friends of the user in the user’s social network, but also can include friends of friends who may be in other social networks as well. That would fit in with the Katango patent Google acquired from Katango, Automated Agent for Social Media Systems which was published (and filed again as a continuation patent) on January 10th.

A screenshot from the acquired Katango patent showing how the Katango social agent works across different social networks.

The patent filing describes how a social media agent can log into a social network for a user, with permission to act as him or her, and collect information about the actions and activities of a user’s connections on those other networks. Might this viral score system collect information about activities on YouTube and other social networks as well?

Calculating a User’s Content Propagation Likelihood

The formula that the patent shows for computing a user’s content propagation likelihood is:

Content Propagation Likelihood =

w.sub.1(“Endorsements”) + w.sub.2(“Shares” & “Reshares”) + w.sub.3(“Comments”) + w.sub.4(“Messages”) + w.sub.5(“Visits”) + w.sub.6(“Purchases”) + w.sub.y (“Search Results Clicks”) where “y” is an arbitrary number, and where adaptive (or adaptable) weighting terms {w.sub.i} are used with each of the various components included in the computation.

Note that these are the same pieces of information that might be collected about the usage of an entity, but in this formula the data is collected for an individual user and people who may be connected to that user (and people who may be connected to them, etc.)

A flowchart from the patent showing how entity usage information might be incorporated into a viral score for users.

Velocity of Propagation

The speed at which an entity, or content item, might be shared and/reshared may also play a role, as an indication of the velocity of propagation of that entity, through the social network and the social networks of the user’s friends.

Different speed thresholds (how much weight the signals might carry) might also depend upon things such as they type of content item involved. For example, video content might be easier to share quickly, and might require higher speed thresholds during sharing and resharing than static images. With a video, one fast threshold might be 100 videos an hour, a medium threshold might be 50 an hour, a slowest third threshold might be 10 an hour. With static images, those thresholds might be 30 an hour, 10 an hour, and 5 an hour on the lowest end.

Weighting Term Based on Social Network Used

The value of sharing and resharing content, in terms of the adaptable weighting terms {w.sub.i} from the formula above may be adapted (e.g., updated) based on the particular social networks in question. For example, a “retweet” (if Twitter were to be included) might be considered an easier share than a status update on LinkedIn – 20 retweets on Twitter might be worth 5 shares on LinkedIn. Those comparative values might be calculated based upon differences in how entities (content items) spread throughout different social networks.

A tweet that I retweeted on Twitter about the business case for clean energy.

Also, the weight attached to the number of visits to a website or physical business location associated with an entity (w.sub.6), may be adapted accordingly in situations where the entity isn’t associated with any particular website or when the physical location of the business associated with an entity is in a different country than a user.

Impact of User’s Taste and Expertise

If someone is known for their taste in photography, and the entity or content item is related to photographs, it’s possible that more of the user’s friends will pay more attention to entities related to photography, and the user may have a higher content propagation likelihood because of their taste in that specific vertical.

Take-aways

The patent provides more details on how a viral score for a user may determine what kinds of ads or other content is shown to him or her, and how that score might be used to match up users with advertisements and other kinds of content.

It’s hard to tell if a system like this would be used with a system like the AdHeat one I mentioned above, or independently of it, or if we will even see ads on a social network such as Google Plus.

The patent doesn’t tell us this, but it is possible that information about the use of entities (content items) might lead to user viral scores that could determine the ads that users are shown while searching in Google itself, or on pages displaying Adsense, at least when they are logged in to Google.

Would Google use such a “content propagation likelihood” to show content other than ads to users as well?

What’s your Google viral score?

I’m guessing that if Google uses a viral score, it’s not going to share the score with us.

Share

37 thoughts on “What’s Your Google Viral Score?”

  1. I only had to read about 1/4 of this post to come to this thought: At some point in the very near future, all content will be some form of advertising (even though in reality we all already know it is just that, right now). And as I myself have discussed about Google+, now that they have all your data from search, browsing, email, and now social networking, they will continue to deliver “ads” based on your profile. What you are discussing here is how the evolution of ads & various other content begins to blur together into just “content”, especially as Google’s other projects, like Project Glass, continues to mature. Hope I’m on to something :)

  2. Hi Eric,

    I think you are on to something there.

    Ads that are actually relevant to your interests, and content that you (and your connections) curate and share and interact with definitely are better than ads that aren’t relevant and have no meaning to you whatsoever. The interesting thing about this is that it considers your activities and what you contribute and remark upon in social networks.

    It looks at a lot more than what you and your friends might have “liked” on Facebook to decide what to show you, like Facebook’s Graph Search might during a Facebook site search. It looks at a lot more than just your latest search query, or last couple of search queries.

    Do advertisements have more relevance when they are based upon implicit queries based upon what you do rather than what you entered into a few fields on a profile? Do they start becoming more content and less advertisement?

    With things like Google Now and Google Field Trip serving content before you ask for it, this could be part of the evolution of advertising from Google. I suspect the intelligent assistant that is Google Now, and the location-based alerts that is Google Fieldtrip will evolve, and will be the kind of thing that gets integrated into Google Glass.

  3. Very interesting analysis Bill. I suppose this is the next logical step for online ad marketing although some of the concepts here such as the speed of sharing are things I hadn’t really given much thought to. I’m just not sure how this will all play out in the end. At the moment I find adwords retargeting to be a pain as those ads keep following me around for ages and I never click on them…I’ve moved on! If then ads appear based on sharing history, what friends of friends have shared, etc, is that really what we want? And content merging with ads, I suppose this happens a lot now anyway but it looks like whatever happens ads are definitely going to become more of a social thing than they are now.

    Anyway, interesting post.

  4. This is interesting, but I’m now concerned about privacy issues as a result of all this discussed here, isnt anyone looking at the privacy issues?
    I mean, day by day people’s privacy is being eroded…

  5. I am still trying to understand the pdfs you have shared but,what I think is Google Adwords is really pathetic.I have tried using it once but I did not get much of the help.Google is better when it comes to analytics and If Google plus starts putting add with the same analytics power it would be much better than Facebook adds.Secondly Facebook adds are not Appealing to me.Size of advertising space has increased and I barely hit any of those adds.
    Google is coming up with many innovative products and I hope advertising space will improve on Google

  6. I like how it would give weight to people who share about an entity that they know about. Seems pretty logical. Not sure how I feel about being part of this global ad campaign, but at least the submitting of personal information collection would be optional. I just hope they don’t keep changing what the default settings are so you have to go fix it constantly as Facebook did with their settings for a while.

  7. Bill absolutley fantastic article!!!!! Extremely in depth and comprehensive, it really made me think. I might add that it’s extremely worrying to think that these search engines are getting this level of information from it’s users.

    It feels like it has a voyeuristic element to it, but on the other hand it’s a fantastic marketing tool for businesses seeking to market their products. It is very obvious these search engines have been using our information for marketing purposes, but your article really made me look deeper into the topic. Thanks Mark, I look forward to reading many more of your posts!!!!

  8. Thanks for this, Bill. Well explained as always.

    One side point, as I pointed out in my post http://www.virante.org/blog/2012/12/21/google-plus-ads-will-there-be-any-what-will-they-look-like/ Google does not have to ever post ads on Google+ itself in order to monetize G+. G+ is aleady monetized by the way it affects your search results when you are logged in and using Search. (Improves them with personalized recommendations, so you like Search and will use it more, thus seeing more ads).

    This patent simply takes that to the next level. My “virality” on Google+ and elsewhere could help Google improve my ad experience on Search (and elsewhere where Google ads already show). Thus my Google+ use is of even more fiscal benefit to Google, without them ever putting an ad on Google+ itself.

    And, as you allude to at one point, this patent could defintiely be incorporated into AuthorRank as well, as one more signal of the authority of an author: his or her “viratlity quotient.”

  9. The power of search engines (mainly Google) nowadays is huge, I think there may come a time when they start personalising results to the point that users are being shown stuff very similar to what they already know/believe, sure this is good but it means that users may not learn new things that they may not have looked into before.
    Added to the privacy issues it may turn into more of an issue that a advantage to users searching and being returned customised results based on the individual

  10. The fact that search engines are now becoming more and more capable of directing traffic based on prior searches is concerning. Recently as I was attempting to optimize my site I found out (probably should have known already) that google geo targets results. Meaning that my location in Chandler Arizona directly impacts the results I see. This is normally a welcome trait of Google but it does make me wonder if there are times where I am misinformed about things I am searching for because of the fact that Google is basically telling me answers rather than allowing the search to be truly natural.

  11. It would be swell if it worked like ads in Gmail, just tag each post with #death and you’d be ad free.

    I’m having difficulty finding the correlation between a “Viral Score” and historical engagement activity though.

  12. “Would Google use such a “content propagation likelihood” to show content other than ads to users as well?”
    I think they already do so. If I search for a special term in a browser I’m logged in, I get other results than in a browser I’m not logged in and I’ve got the feeling, that this difference is sometimes based on the Sites/Profiles I’ve added to my Google+ circles.

  13. Sounds like a Viral Infection to me! I am one who prefers to see content that is less personalized. My concern is that things like polarization of politics is just further fueled by people being only presented with information that an algorithm decides is best for them. It further makes it difficult for one estimate what other people are reading to form their own opinions. Finally, some people may be manipulated with information without the knowledge of the public at large. I don’t like this stuff!

  14. Bill,

    I’m very much a sceptic/laggard when it comes to the over-enthusiasm I see in the promotion I see for social media. However, the sheer fact that Google+ exists implies that it should be taken seriously. Second, the fact that Facebook sees it (Google+) as a credible competitor implies it is a threat to their business model. And thirdly, as mentioned above, advertising is the de facto funding and profit stream on the Internet. Therefore, it seems likely Google+ will run advertising.

    The counter argument is that Google has such deep pockets that it could run a rival social network to Facebook et al without the need for the intrusiveness of advertising – particularly the behaviourally targeted variety.

  15. Bill,

    Loved the specifics of this post. Advertising is at the core of the G dynasty. It fuels the fire and greases the wheels of their runaway freight train. The problem that exists for Google is that they have been unable to crack the social nut. (I’m not saying that research scientists aren’t social.) LOL Just not in the way that the population at large is. The more aggregated data they receive, obviously, the better results and Thought Police Ads they will serve. I have a problem with my search engine being involved so deeply with advertising however. Kinda like Arther Anderson accounting and auditing Enron…. Keep them flamethrowers coming Bill. Great Stuff

    “Might this viral score system collect information about activities on YouTube and other social networks as well?” You betcha. Especially all the channels in the Google stable.

  16. Can we connect this Google viral score to the search results? I mean one day it will also influence search results as today backlinks are influencing. But than I think that if this happens than it will become more difficult for the new website’s to sustain. I have one more question, do you know when Google PR update is going to happen?

    Thanks for sharing this post.

  17. I wonder if Google is aiming to sell segmented users to advertisers at a premium based on their likelihood to share content?

  18. how is it that yesterday i am talking to @tonyverre about work related Google stuff.. and i say.. “who knows” i bet slawski will have a blog post about it…

    btw.. thanks for blog post..

  19. I have read a bit about how users interact with ads on social media sites and currently people are happy to “like” certain brands or products but they don’t buy anything via social media sites because they don’t use sites like G+ or Facebook to buy stuff. It looks like Google are looking for ways to display content or advertising in a way that engages people differently.

  20. Very interesting well researched investigative post.

    For me, it will make a lot of sense for Google to implement this kind of social network viral scoring for ads on Google+. They clearly want Google+ to succeed and social networks are build on content shared by its users. By using an ad-serving algorithm specifically aligned to the purposes of sharing and re-sharing, Google is strengthening the value of the Google+ network by providing more shareable content directly to users.

  21. From what I’ve read of that … they could utilise it to decide what Adverts to display in SERPs based on SocialMedia activities for logged in/recognised users.

    Alternatively, would that not be a viable/effective method of identifying people/pages/communities/events to suggest/recommend?

  22. Hi Lyndon NA,

    I think that might be what we see, social influenced advertisements, but in search results.

    And this process would be one that could be used for recommendations and suggestions on social activities, and what’s “hot.”

  23. I think it might be the reverse – rather than all content becoming ads, all ads become content. “Content marketing” is the rage right now, and viral videos are effectively content. It’s in Google, Facebook etc’s interests to have “ad units” propagate, so it makes sense that they’d be working on mechanisms to make that likely to happen.

    High production value video and long-tail highly relevant content marketing seem to be two potential winners here.

  24. Hi Sean,

    Google is exploring how they might have both advertisements and content be seen by people who are more likely to be interested in them. Of course, at this point we’re talking about a patent rather than something that has actually been put into place, so we can’t be certain that what is in the patent will become a business practice.

    I think an ad is still an ad, no matter how you dress it up, or how finely produced it might be. I don’t think we’re going to see informercials become content on a successful and continuing basis. In the long run, having a Google move down that direction is more of a risk than a benefit. It has the potential to betray the trust of people who look for information through what they perceive to be an objective resource.

    Relevent ads do have value if they are something that have the potential to help people to view them. But I think if it gets to the point were the ads can’t be distinquished from the content, that could be a real problem.

  25. Fascinating article, Bill. Google knows that it must continue to be relevant and keep a step ahead of the other engines in order to keep dominating the market. Whether or not their moves actually cause them to lose market share is yet to be seen. But there’s no question that they want to be the main source of marketing for businesses tomorrow.

  26. Hi David,

    Thanks. Google does seem to want to keep on innovating and staying a step ahead if they can. They are forging ahead with both social features and knowledge base information included in search results.

  27. I like how it would give weight to people who share about an entity that they know about. Seems pretty logical. Not sure how I feel about being part of this global ad campaign, but at least the submitting of personal information collection would be optional. I just hope they don’t keep changing what the default settings are so you have to go fix it constantly as Facebook did with their settings for a while.

  28. Hi Bolocan,

    The patent notes that while this could be used for advertising, it could also be used just to show people things that might be related to what they’ve shown their interests to be. As for “changing the default settings,” I’m not sure hat you’re talking about.I would guess that it’s highly likely that if you show interests in different things over time, that whatever your viral score might be would change.

  29. I read someone talk about privacy issues. People need to understand that whatever is posted on the internet, by themselves, is in the public’s domain. You cannot post up something, and then cry about privacy afterwards.
    On the other hand, if someone posts about someone else, and that someone feels victimized in some way, then it becomes a “privacy” issue. Another comment I read, mentioned everything on the web being some form of “advertising campaign”. I totally agree with this. The web is being transformed into just that!

    Chris

  30. So if I read this right, Google is building a Klout score for Google+ that it will use internally for determining whether content will be displayed. I can hear the SEO gamers wringing their hands already on how to manipulate this algorithm to make poor content “go viral”. #Blech!

    Great article, Bill! I do hope that this isn’t just another score like Google PR that will become the basis for manipulation, as opposed to publishing and distributing quality content.

    Cheers!

    –Sean

  31. Hi Sean,

    Thanks for asking some of the tough questions.

    I don’t think that Google’s intent was to build a Klout score, but rather an algorithm that might actually be useful to rank content. At this point, the Klout score really exists more for for entertainment purposes than anything. Part of the reason that Google is unlikely to reveal exactly how they might use this score, if they decide to, is that they do want to avoid having it gamed.

    But the things that are revealed in the patent actually involve a fair amount of actual work, from what appears to be real people. The great thing about this score is that it could potentially take a lot more work to try to “game” this kind of score than it would to actually get real people to build a social score for themselves. Search engines are attempting to make it harder for people to spam the systems they create, and one of the ways to do that is to force people to attempt to rank well legitimately. All ranking signals are targets for manipulation, but if search engines can make them harder and more costly to manipulate, the winners are searchers, and the people who strive to provide them with value.

  32. Very interesting well researched investigative post.

    For me, it will make a lot of sense for Google to implement this kind of social network viral scoring for ads on Google+. They clearly want Google+ to succeed and social networks are build on content shared by its users. By using an ad-serving algorithm specifically aligned to the purposes of sharing and re-sharing, Google is strengthening the value of the Google+ network by providing more shareable content directly to users.

  33. Thanks for the interesting post Bill…. Thought I could just glance through it and in the end, an hour has passed me by. Is there any stopping Google?…. Didn’t they once say that about Yahoo and MySpace? (Have you seen the new MySpace by the way…. pretty good).

  34. Interesting read and having only just registered with Google+ am stumbling my way around it! Look forward to reading more, now that I have discovered your site.
    Alison – over in Australia :-)

Comments are closed.