Monday, July 23, 2012

// // Leave a Comment

Issues with Like-Dislike Voting in Web 2.0 and Social Media, and Various Defects in Social Ranking and Rating Systems - Confused and Vague Design and Measure - Psychology of the Crowds - Corrupted Society Preferences and Suggestions. In Facebook, Youtube, Twitter, TV Networks...

Issues with Like/Dislike Voting in Web 2.0 and Social Media, and Various Defects in Social Ranking and Rating Systems

  • Confused and Vague Design and Measure (copying an aggregation mechanism in brain)
  • Psychology of the Crowds - Corrupted Society Preferences and Suggestions
In Facebook, Youtube, Television, Twitter...

Simple numerical ranks are the easy way to quickly see what's "important", "good", "of high quality", "interesting", "impressive", and the most real measure - popular. I'm not talking about the Google's Page-rank, which is somewhat "objective" regarding the popularity of a page on the web.

I'm talking about the votes by users. In the past, Youtube and other sites had "stars" (from 1 to 5) or other "ratings", perhaps suggested by the PageRank algorithm or something. How do you measure and decide that it deserves exactly "4 stars" or "5 stars", or 3 or 2, what exactly the random viewers compare to other publications?

Then the trend changed, and Youtube and many other blogs and content-sharing and user-generated sites switched to "Thumbs up", "Thumbs Down", "Like", "Dislike", as indicators of public appreciation or interest, in order to simplify it and perhaps to make it more "objective".

What exactly is interesting, what does it mean to be interesting for each of the ones who liked/said it was interesting, and it's interesting for whom? 
OK, so you claim you liked "this" video, or you liked "that" publication in Facebook or Twitter.

Great - and what do you mean by "I like this"?  What does "this" mean, and what does "like" mean? The same question goes in psychological context when somebody likes you or when you like somebody. You don't like "everything" in her, you like particular details of her, her behavior to you, etc.  but the overall feeling of the "total pleasure"* is somewhat a "generalizing" measure of all kinds of "likes", especially when one fails in distinguishing and reflecting on his own preferences.

Simple Neurological Basis of Like/Dislike

"Like"/"Dislike", the aggregated "overall feeling of "total pleasure" is in fact the concentration of dopamine, endorphines, oxitocine, serotonin etc. in particular parts of the brain in a moment when the conscious part of the mind, high-level in the cognitive hierarchy and abstraction** is focused on particular "objects", so it makes a connection and infers that "this" thing that fills the attention up is responsible for the overall pleasure, and the mind likes "this" thing.

Another reason for the like/dislike aggregation are perhaps the severe output bandwidth limitations of the brain. Complex sensations, memories and reasons are hard to explain and express in details.

Like/Dislike with no particular reason is the last resort of mind when it can't understand itself. One would justify what he does by:

-- "I do it, because I like it", i.e. "it makes me feel good",

which is similar to

-- "I do it, because I want it."

Still "it", "this" and so on are vague - "it" is what is in the focus of the mind when the mind declares this state.

For a meaningful measure it's important what exactly was "this" and "that" that you liked, and what exactly you did like, because there are completely different reasons for one or another person to like or to dislike one video or another, and also there are "noisy" social reasons making popular things even more popular for free thus polluting the big picture. All of those reasons make that  measure vague.


  *Some would even call those physiological processes "happiness".
**See the work of Todor about the sensorimotor generalizing hierarchies.

The article below explains some observations regarding the confusion of the Like/Dislike rating, and why it's vague and corrupted. The work doesn't pretend to be completely exhaustive, more details and other structured classifications and laws might be extended later.

OK, now you clicked "LIKE" for that clip,
but what exactly did you like and why?

  • The topic,  the "substance" of the performance and the skills and talent depicted?
MTB bikers would like MTB clips, bidding guitarists would like other guitarists' performance clips - such as the "Canon Rock" phenomenon. Indeed, this phenomenon happened in the young months and the beginning of Youtube, when the first clips in particular topic have gotten the market. This particular performance has frequently appeared on top in queries and there was not a good competition in the niche, such as "electric guitar", "guitar playing" etc.
  •  The quality of the performance? 
In case of music, tricks, acting... Or some aspect of the performance that you like.
  •  The technical pixel quality of the video?
Expensive cameras, good lighting; frame composition, smooth or complex moves of of the camera, high-speed cameras? If you're into these or care about this, or if you're not into this and you're impressed.
  • The technical quality of sound or music?
Recording equipment, mixing, pleasant sounds (for you).

  • The appearance of the performers or a particular participant in the clip?
You like particular actor or actress, or director or musician. Or you like the personal attractiveness of a performer who's new for you, no matter the rest of the components.

Hot girls and "hot clips"are liked even if they just walk or open their legs (even not naked). So you liked "this", in fact you liked this feeling, and act as you'd have acted if you have met such a girl in a close contact - with a compliment, a smile etc.
  • The location
This is a video of a city/place I live in, or I've been, have sentimental memories, etc., "I like it" - the location. No matter how bad it's shot, or what's on the video.
  • The content of the "political" message - you  agreed with the author or the "cause" or you identify with it
This video supports my views, I like the message, or dislike when you disagree. Even though the video is perfectly directed, shot, presented; has perfect acting, technical qualities etc., if one disagrees with the political message, he'd dislike - the "cause".

Boring or naive badly shot clips, depicting a "nobel" cause may be liked by ones who want to identify with such causes and declare they are with them.

  • Support your peer, friend, fellow countryman etc.
This is a video of my friend etc."This" is my friend... I want to give him a hand, make him famous. I'm proud he does this etc., or I also helped him.

A video depicting in a "socially positive way" a talented one from the country X (for example in a reality show that "searches for talents") would get support from thousands or millions of fellow-country men, because of their attempt to "show to the world" and make it important, and because "it" makes them feel  pride. It's the pride that they like - even if the essence/substance of the performance might not be that talented or extraordinary.

That's why web voting is intrinsically corrupted. Nations and social circles who are not ethical or who are just more or are more "fanatical" about displaying "how many they are", "how great their fellow country-man/team-mate/football club" etc. are, would vote more and would distort the picture.

It's not the greatest, the best in the art etc. who wins - it's the one who is somehow connected with more persons who're committed.

It may be claimed, that this displays "social significance" - I myself would say that it rather displays the status-quo. A talented one who has not many peers yet, may be socially suppressed and isolated by mediocre average-Joes who however are more "social", have "networking skills" and are generally more similar and close to the crowd which supports them, i.e. they are more like the crowd that likes them.

  • Support an author or someone who participated in the video
  • Indicate that you belong or want to belong to "this" (pick any)

  • You were surprised (positively - like) or you didn't see what you expected/wanted (dislike)

  • You were amused - doesn't matter why

  • You were impressed by some extremes in the video - "the best in the world", "the fastest", "the dumbest", etc. that's often used in the titles of videos to attract random attention

  • You did understand the message (like) or you didn't understand the message, you feel "dumb"so you dislike and comment "dull, boring, ..."

  • Mark the clip in your playlist (in order to show it to somebody else later)

    • It was useful - reward it for serving you a purpose

    Searching for a commercial or amateur or community review of a product, saw the review, found info that you wanted, etc.

    • You liked the product, depicted in a marketing video

     You found a product by browsing and liked it. So "this" is the product you found, the quality of the video etc. doesn't matter.
    •  Overall, all of it, more of "this"or some of it?
    In fact everything may be reduced into the "feeling" that the viewer gets when perceiving the video, and if it passes a threshold to make her click a button. See the introduction above and see also below. The mixture of all this makes "like" fuzzy and confusing.

    It displays a vague "social support" of "this".

    Try to define "social" clearly, it's a messy meaning-all word alone:

    - Wherever more than one people are involved?
    - What's "useful for the society" - what's society? The ones who have the highest number? The highest force? Wealth? Power? The ones who are the most and the ones with power have brainwashed to follow?  Etc.

    • The "social support" part and ranking systems confuse this measure more
    The most famous (visited) or liked clips or publications at a given moment go on top in rankings and are promoted in lists.

    Many random viewers click them, because they are on top, not because they care or are interested preliminary - that's due to the tendency of people to pick:

    -- The options which are on top, i.e. need little time for search, i.e. need the least time for search and get, the shortest attention span for acquiring, the least processing requirements
    -- The options which are "socially promoted", "nobel"  (humans are trained for this), or sometimes exactly opposite - "bad" (now a "revolt" against what people are trained as "socially correct"), like something nasty.

    If a celebrity posted a video of her walking in the bathroom, showing her p* or d* or produce a storyless short film with her, the video is likely to get viral and would be liked a lot.

    If Jon Doe from Nowheresville does the same thing and publish the same short film - nobody would even notice.

    It's not about the content, but the "social importance", the "crowd importance" and overall - marketing and advertisment, which is close to "distribution" in more physical terms.

    Some producers of anything are not as popular as they are just because their goods are technically or whatever better than the competition. They just have better distribution - of course if their product is in all shops all around, there will be more people buying it.
    AMD vs Intel and Coca Cola vs Pepsi
    There was such a condition in the PC CPU market - AMD had a time when they produced superior processors than Intel, it was the era of Athlon, then Athlon 64 and many of the versions of Pentium 4 starting about a decade ago. Pentium 4 initially was a flop, slower than Pentium III in clock-by-clock dog-fights, employing initially only a very expensive RAMBUS RDRAM memory. AMD's CPUs were way better in games and gave much more performance at a lower price. Every smart consumer  was purchasing Athlon in that time. AMD first introduced 64-bit x86 CPUs and multi-core, their Ahtlon FX CPUs were leaders in server applications and the fastest CPUs. Intel needed years to get the crown back with the Core CPUs, which inherited the Pentium III line.

    AMD had a rise in their share during that period, however AMD was and has ever been too small a company to beat Intel, as of 2012 data it's like 1:10 in terms of human power. Even if there's very high demand, AMD couldn't supply enough of CPUs to flood the market like Intel has been doing, and Intel has been selling a higher number of technically worse CPUs than their technical and price qualities would have deserved, if Intel didn't have so much bigger production and distribution network.

    More sales lead to more revenues, but it requires too much more revenues to beat someone who's 10 times bigger than you. Not to mention the dirty contracts with computer vendors which someone that bigger can do and was found to do, search to see the cases yourself.

    Another story is Coca Cola with their famous unfair contract with vendors that they shouldn't sell competitive drinks in the same rack/shop (at least in Bulgaria, it's a well known rule). The shop should choose either Coca Cola or Pepsy - the consumer has no choice. Coca Cola is sold in every little village around Bulgaria. Sorry, I don't want any of those, but that's another distribution dirty trick that makes a product appear more "popular" or "liked" while it's just pushed-in-your face harder than the product of the competition. 

    The average consumers just take what's pushed harder into their faces, not what is "better".

    In this case "pushing the thing into your face harder" is what "it" in "LIKE" means, it's rather a measure of how the marketing forces are "pushing", not a measure of the opinion of the users.

    The Already-Being-Celebrities are advertised in mass media for prolonged times in many media. As a product they have good  "distribution", their names are queried in search engines, cited in articles, queried together with other celebrities, and the avalanches about their name grows.

    Not all viewers click like or dislike, only ones who have predisposition to do it

      They seem to be either ones who are:
       - Overconfident in their opinion
       - Believe their opinion matters
       - Want to change the world/make a difference
      - Ones who have strong enough feelings, provoked during watching the video (or at that very moment, they might be affected by something else)
       - Ones who want to be "connected" or to identify - 100000 people liked it before, there are 1000 comments - I also have an opinion! - exactly the opposite logic also applies here for another part of the viewers - some of them would not leave a like or a comment, because they'll see it's meaningless and useless already
      - Some users mark their reaction only if they don't like the video, to "punish" the author, but they don't like if they thought the video was fine or they liked it
      - There's an opposite group - some users mark their reaction only if they liked the video or have something positive to say, because they think it's impolite to insult people like this

      Regarding commenting:

      - People who are slower typers apparently are much less likely to comment or their comments are likely to be very short and meaningless
      - Regarding the most comments that appear on top at the mass video sites, the most of the ones who comment are ones who want to say something bad, offending, and often are children or teenagers, saying something obvious or something meaningless that another 1000 people has said already (in those mass publications) like "Wonderful", "Great", "Awful", "Cool", "This is creepy", ...

      In general commenting in a mass thread is meaningless in practical terms, unless you're a... much faster and smarter than a human, i.e. - a thinking machine. Practically nobody would scan or read the comments if they are more than a few (one page, rarely more, if one is doing a research or something). Also unless the comments are very short (a line or a few), a few people would read them all or answer meaningfully - it's hard for people to focus, and they hardly care that much. In fact that's not for a bad reason - it's apparently hard to find a good reason to focus on those random comments by people who're dispersed around the world, don't know each other, don't care about each other etc - that's purposeless communication that don't create new bonds and friendships, just wastes time.

      Sure, the authors may find useful statistical information in the comments if they are "en mass".

      Number of viewers and the comments

      The number of previous viewers affects the number of future viewers, it's even more evident about the number of comments. A video with many comments and views is more likely to be commented more ("that's something important, I also have an opinion!"), than a video with no comments.

      This has to do with the psychology with the age and social groups which comment the most.

      Comments and likes would be more meaningful if everybody voted, and the votes were way more specific on every single component, and not about "this" video meaning possibly 100000 different details*

      * Yes, people have largely the same minds and opinions as a group they belong, depending on their social status, education, peers, media accessed etc., and they are people from the same group are likely to like or dislike mostly the same particular components. Such selective like/dislike procedure would illustrate those  groups. However, as mentioned in the introduction, the severe and pathetic limitations of the output bandwidth of human brain (we express information miserably slow)  is something that would make most people deny marking their opinion more specifically - too much time, too much efforts, and nobody really cares about those "details". That's another fault of brain - average people don't care about the "details", they are interested in vague overall "good/bad", "like/dislike", which easily direct their behavior or choices.

      Of course that's exploited in the market, everywhere if there are some "staggering numbers" which are technically objective that may be used to impress users as "better than the competition"who don't understand and don't care about the details. In the digital camera world, it was the "Megapixel war" - more and more fake noisy pixels in consumer cameras, and people buying cameras that take worse pictures but has a bigger number, printed on the body.  In the CPUs it was of course the MHz, GHz, number of cores. There were also "ratings" and of course benchmarks and detailed reviews, but there are millions of non-informed users who can't focus and are not willing to understand on those "hard details" and wouldn't decide on them, they rather need just two numbers to do the most elementary "bigger/smaller" comparison.

      Number of real committed viewers is less than the counters due to short attention span and other stimuli which compete for the viewers attention

      The raw numbers displayed as statistics are surely significantly higher than the "real ones". The longer the video, the less viewers would have patience to watch it whole. A lot of viewings are just hits of the "play" button, because the video is displayed on top, as watched, famous, etc.

      Be the first in finding a niche for something that many would look for - get viral. Once got viral -  you'll get even more viral!

      Publications which are first in a niche, get easily viral, like it happen with a new products in the market, because there is small or lacking competition in the start, and the popularity of the pioneer feeds its future popularity and fights the new candidates. If you check the date of the original JerryC "Canon Rock", it's from the early 2006. Youtube was created in 2005, but got really popular in 2006, and that famous "phenomenon" happened - it just filled the niche of domestic guitar players trying to find audience and participate in something together.

      Trendy videos get even more trendy, because they pop on the top of the lists with suggested or trendy items, and viewers get their need to watching "something like this" satisfied by what's given to them there. For example the "Fial-blogs", "Fail compilations" etc. Any of many would serve, but the ones who pop on and gain enough views would keep and save their rank. The same effect happen in sites for jokes and in facebook in their sub-communities.

      • The Rating and viewers number and appreciation depend dramatically more on the general popularity/rank of the source/media, than the quality of the show/product being aired itself

      I saw that effect in Bulgarian TV market, I guess it applies everywhere. I would generalize it like the following:

      It doesn't matter what is aired in prime-time in the most-popular television.
      It will get the highest rating (according to the rating agencies). And no, the high rating is not because the television producers "has skills to recognize quality products" so that they select only the "best shows"for the prime-time.

      It's just the time when the most people are on the TV, and the most popular television gets the most viewers in that time-slot. People would watch anything you show them. Why? My general explanation: it appears because they are lazy to choose and pick themselves and they get satisfied by whatever.* (See note)

      • Rating going down from a million to a thousand

      There is another illustrative case for that "law" that I've noticed in the Bulgarian TV market. An amusing and high-rating satirical TV show about funny mistakes on the TV channels was suspended due to a scandal and, the TV claimed, "low rating".

      However the producers web site, which was advertised in every show and is very easy to remember, continued working. The producer claimed, in rage, that they will keep their show even if they don't find another host TV (eventually they returned to their previous host TV...). I was curious to see evidences for my prediction that there won't be many viewers online, even though there was a lot of noise about them during the scandal, includes on the TV that they were leaving and is in the top of the ratings.

      The prediction was true - they've been publishing new content, unpublished on TV for one reason or another, but the videos seem to have on average the miserable 1000-2000 views for several weeks or a whole month of exposure, some 3 month-old videos had maximum of 4000-5000.

      Keep in mind that in their best on air they have had probably 1 million or even more  for each of the shows, and 500-750K in not that good times (of course if the rating agencies could be trusted). Also they have massive "celebrity advantage" and advertising - it was a very funny show, lately hosted in the most popular TV (previously in the second one), presented by some of the most popular Bulgarian comedy actors and integrating hilarious material and characters from the entire Bulgarian TV market.

      That means: If you don't push it into the viewers faces (on the TV, one or two buttons away), only the very dedicated fans would even care about your existence. The crowd wouldn't care whether you existed or not. Viewers' attention would be immediately grabbed by something else.

      * People would watch anything you show them.*  -- Sure, it's not absolutely literary, but close. For the Bulgarian TV market, an example - technically and artistically superior soap opera (telling that as an independent filmmaker and an actor) had less of rating than another soap opera aired before (but both were very high though).

      The higher rating one however depicted the world in more every-day like over-acted, and melodramatically over-the-top and protracted slow fashion. The slightly less-rated production had vastly higher cinematic standards, action, camera and impressively convincing acting, while the higher rated one had mediocre camera-directing work for such a big production, mediocre and banal story and storytelling, preaching and cliché style, but the the mass soap-opera viewers liked it. Apparently they couldn't and don't recognize those aspects as faults - their "it" regarding their "like" is something else, it's that "overall feeling" they get or their wish to be part of the society that watches it and empathise with those cliché stories.

      There was another Bulgarian soap opera, which in my opinion is artistically a disaster, awful acting with inappropriate face-expressions, dull dialogs and transparent script. The mass audience didn't care - the worse, it found it interesting. I'd be afraid to share my opinion about the film, because the crowd, apparently very highly educated cinematically, would tell me how "dumb" I am and how I don't understand nothing about movie making. However, this disastrously bad, in my honest opinion, soap opera was written by a famous Bulgarian writer and aired in prime time in the most popular TV (according the rating-agencies and advertisers). Yes - it had the top rating, regarding the rating agencies, in the end of the first season they claim 1,2 million viewers, 47% of the market.

      In the same period another Bulgarian-made TV series, action-drama, aired on the third-rated television in Bulgaria, produced with very high cinematic standards had a fraction of this, a half or less in its best, generally a few times or many times less.  The producer claim they have been selling it on the USA and Asian market.

      Even though technically and artistically it was in another dimension compared to the regular banal soap-operas, its rating was a fraction of the rating of them, aired on higher-rating televisions - the rating mirrored the general  ratings of the televisions and was predictable by it, that series was on the 3-rd rated television.

      The "prime-time" slot means for the crowd audience "important", the audience talk about it and makes it more popular, no matter what's on prime time


      "Public involvement" makes "it" important

      Not that's an invention to say, but you can hardly get 1 million views of a video depicting how to program device drivers or how to operate an electronic multimeter, or how to code a compiler* - for a video about compilers 10000 views and 10 likes might be very high in its small market place. Of course a video that requires specific knowledge, expertise or attitude in order to be enjoyed, appreciated or understood, would not seem interesting to random audience, and it would be disliked and marked as "boring", because it bored the inappropriate audience.

      The comparison of the popularity/numbers of viewers or so are meaningful only in the right "classes"/categories - similar types of videos with similar degree of advertising etc. Of course something that's promoted in all kinds of media and is generally more popular/accessible to anyone - such as pop music/the "idol" performers - will be more popular than a clip of an underground school band, which has no advertising (distribution).

      The numbers reflects the general public interest, the number of interested persons, and the number of persons whose attention was somehow attracted to this video/publications from particular topics/sites/media (see above the ranking lists, advertising and celebrity status). In order to find the position of a particular video in the big picture, it should be compared to similar videos in its popularity, accessibility and distribution class. Lady Gaga should be compared with Madonna, Rihana etc.

      Some videos depicting mediocre or everyday life stuff of which anybody is interested get extremely popular,   when catching a niche - you certainly have seen such examples on youtube, I want promote them though.

      * In a linked sequence of clips - tutorials, documentaries, lectures etc. - the higher the number of the episode of - the less the views/audience

      This is apparent in most of such series, for example I saw it clearly in one of the now trendy online courses, the AI-course from Stanford which I checked a long time after it has finished, as a "guest". They claimed that 100K+ (don't remember the exact number) of students registered, however there's a classical diminishing pattern of views in their lecture videos on youtube - which are accessible without registering for the course.

      In the latest topics, supposed to be the most interesting, the videos from the course had just a few hundreds of views, some of the homeworks have the ridiculous TENTHS of views for a global, world-wide course, lead by one of the most famous authors in AI and Google's research director or so, and one of the most famous experts in computer vision - who was head of the team that won DARPA Grand Challenge for Stanford and created the Google's self-driving car... Besides he happens to be a talented and amusing teacher.

      So if these results are a guide, if you produce a scientific cutting-edge R&D lecture video and got 50 or 100 views - don't be ashamed of yourself, it's a decent popularity, the problem is in the world we live in, not in you.

      The prevalent "public opinion" suppresses the opinion of ones who disagree

      Surely some of the ones who disagree (who "unlike" on publications with many "likes") do it in order to display that they are different (not because of some personal opinion), but often on youtube there are top comments saying that "132 people can't play the drums", or "121211 people are jealous" or "dumb" etc., which are liked by the prevalent ones, the majority.

      Generally previous likes or dislikes and comments suppress the opinion of the future ones.

      If all others like something, you don't have right to dislike it, according to the "society", "there's something wrong with you" - sheep syndrome, the power and the psychology of the crowds. Individuals feel stronger in their opinion, if 100000 people think like them - therefore they are right, they conclude. It doesn't matter if the 1 who disagree is smarter, more experienced or just have the same rights as them to like or dislike. They are the mass, the power, so they believe the right is there.

      Going into politics, that's a big bug in democracy. Giving the full government to the ones elected by 50.0001% of the voters against 49.9999% of the voters is not fair, it's a dictate of a 1/2 over the another 1/2.

      50% of the voters do not agree and don't want that government - only a jointly government would be fair.

      Howver, it's also the case if the difference is bigger, even if it's 90-to-10 - who decided that it's fair the ones who are more to decide from-then-on for the life of the rest and that's "fair"?

      It's the raw force of the numbers that win, and the social status-quo which has established those rules in the past, and once established, those can't be changed against the status-quo and the power of the ones who enjoy or exploit this situation, not without something radical.

      How the votes are collected and are the votes equal is another issue, the law says people vote is equal, but is someone with IQ 70 as aware of the politics as someone with IQ 170, and one who has a preset "opinion" and doesn't care for details, against someone who's an expert and curious. Not that the choice of any of them is "correct", but who of them is easier to manipulate, thus corrupting the system? The ones who manipulate better - win in such conditions.

      Society and social number-based ranking methodologies like all above are confused, not fair and corrupted like the ratings of the clips and publications in the social media and on TV...

      To be continued...

      (C) Todor "Tosh/Twenkid" Arnaudov, first published on 23/7/2012 in "Todor Arnaudov's Researchers".

      An edition from 24/7/2012 - extended with several paragraphs, mostly starting from "Suggestion:
      Comments and likes would be more meaningful if everybody

      I am searching for research and business partners, and any supporters and friends for my projects.
      Please check:
       and the rest of the blog, if you... liked this one. ;)

      An edition from 10/9/2012 - corrections of typos and other minor editions

      * 17.12.2023: I noticed the blogpost title was "vaguge" instead of vague.

      0 коментара: