Tag Archives: video

Russian Chat Roulette

One of the latest viral internet sensations is ChatRoulette, which is a combination of "Hot or Not" where you are shown random images and ICQ where you could have find random users to chat with. The ChatRoulette website is so new that it barely has many of the basics in place such as a logo, user logins, search filters or any real censorship options.  A secondary issue is that much of the media coverage is ranking better than the actual website they are covering, which same issue had caused Facebook login issues last week.According to Andrey who responded to the NYTimes questions sent to the website contact email address, he answered a number of questions such as it was just the 4 Google Adsense Link ads down the bottom shown in the image in red that now can cover all the websites expenses but can it be that simple?

Looking into some of the claims of Andrey, the servers described appear to match their german hosts NetDirekt's products and the domain was registered by Moscow based RU. The biggest point is that maybe the NYTimes made a mistake in the translation as it seems although currently NetDirekt's products can handle the website traffic is the figure printed real? I can see the websites rapidly increasing cost may cripple the project and cause a decrease in service quality similar to the Twitter FailWhale and the service may not be sustainable long term without more significant investment.  As based on the NY Times article the figures translated ChatRoulette was using around 7GB/second or 604TB/day or 18396TB/month...I have estimated if they were to switch to one of the cloud based hosting providers such as Amazon EC2 the monthly costs would be around US$1.47 million dollars. I'm not sure AdSense would cover that much bandwidth costs each month, as the growth rates are extermely high and not likely to be sustained without more features and improvements to the interface, and they should consider a solution such as CloudSplit to ensure can manage their hosting costs in real-time.

The interesting stats for the website is that for a service based and started in Russia it is very successful internationally but also means that language support and local censorship laws and regulations will start to impact its service, the Alexa.com estimates traffic by country:
  1. USA (26.1%)
  2. China (17.9%)
  3. UK (5.6%)
  4. Turkey (5.3%)
  5. Italy (5.2%)
  6. France (5%)
  7. Spain (4.5%)
  8. Germany (4.2%)
  9. Brazil (4%)
  10. Holland (3%)
The audience demographics compared to the general internet populat show a larger younger segement of ages 18-24 and much lower number visitors over 35 than other websites generally attract. The gender shows a massive skew towards male visitors and our quick test showed us that many of the random connections were males, with a high portion of the 18-24 demographic. The image below shows some more detail on how Chatroulette compares, a worrying factor is the high portion of school/home based users who maybe exposed to unwanted attention or images.

Is it the Next Facebook?It is too early to call this the next Facebook, but it could easily be expanded and developed into a very interesting Facebook application that will increase its audience and possible revenue sources but not before censorship and user registration is enforced to some level.  The click-stream shows that although Google is one of the biggest sources of visitors, Facebook benefits by attracting 15% of outbound clicks. This is likely users are sharing Facebook links to profiles, applications and fan pages. One of the key factors that gives social websites its power is the increasing time spent on site or engagement but a bad sign is that ChatRoulette's users are spending less and less time on site as shown in the Alexa chart below.

If you want to see some of the more interesting /disturbing chat sessions that have been covered check out rosielikescats photostream, a word of warning about what the images contain. It seems to be a product that may have some potential but needs a lot more investment and development to ensure it continues to grow its user base. A likely step would be to implement a user login that can atleast start to allow users to filter their random chats requests and block abusive members. Give it a try and just a word of warning be prepared to see and hear anything and I warn you that Chat Roulette is addictive....April Update: Chat Roulette continues to attract a lot of attention from people search out more information about it, so does that mean its still growing?

Related Images:

Yahoo patents video CAPTCHA

Yahoo's video CAPTCHA patent generally revolves around the central idea "The invention relates generally to computer systems, and more particularly to an improved system and method using a streaming captcha for online verification."The video CAPTCHA patent appears to have come out of Tel Aviv University's Blavatnik School of Computer Sciences, who advise is purely research but it encourages a new way of thinking.While i think this is a great move forward to combat spam and reduce the effects of social engineering but as most people seem to hate and often find some of the CAPTCHA's difficult aren't they just making it harder?I agree that one of the central issues is bandwidth and it is something that really hasn't been thought through with many of these patents lodged by tech companies.  Unless they plan to ignore the fastest growing segment "mobile" which is already having bandwidth issues via AT&T in the USA.The only point I could speculate is that it might be used for other devices such as Satellite TV, public Kiosks or secure websites such as Banks which seem suitable platforms for this technology as they maybe inclined to be more open to social engineering.It could be more useful to ensure that if someone is watching over your shoulder what you enter, the video patterns may make it harder to read and even impossible to recall? To reduce the chances that a non-human is trying to guess the characters the patent allows for the response to be timed, which can further impact those with slow internet connections.Emergence is impossible for BotsThe next steps revolve around 3-D images but the first task is that they have to discover the parameters of "perception difficulty level" to ensure the application can be used for future security applications. The process is focused around how humans can perceive emergence images or the ability to aggregate information from seemingly meaningless images. The image below is such an image that we can process but computers will fail to understand.Emergence Image TeaserIf you look closely at the image is there anything that hops out....Another one of the sample images based around this is more of a best friend to most men.So why not just improve KittenAuth?So while this is a great move forward to prevent spam comments it seems to be a more advanced version of KittenAuth which requires visitors to count the correct number of animals in the image. The problem was that KittenAuth can be beaten with brute force attempts and requires a massive database of images. With time the bots can begin to learn so the image set may have to alternate between animals and objects and show below.

The other item identified in the creators improving KittenAuth post was that the bots would take screenshots of the CAPTCHA and a human would enter the relevant code which was passed back to the bot.  The Yahoo patent timing element would likely make this process much harder or atleast require more resources and make the site less attractive to spammers.

Watch the future of CAPTCHA

If you want to watch the video of the likely spam benefits of emergence images and how algorithmic trackers struggle to capture anything more than garbage. This makes it possible for the same image or video used more than once as slight elements such as colours, size and length could trick bots.
http://aiw2.uspto.gov/.aiw?Docid=20090328163&homeurl=http%3A%2F%2Fappft1.uspto.gov%2Fnetacgi%2Fnph-Parser%3FSect1%3DPTO2%2526Sect2%3DHITOFF%2526u%3D%2Fnetahtml%2FPTO%2Fsearch-adv.html%2526r%3D4%2526p%3D1%2526f%3DG%2526l%3D50%2526d%3DPG01%2526S1%3Dyahoo.AS.%2526OS%3D%2526RS%3D&PageNum=&Rtype=&SectionNum=&idkey=6FC696AE4CAE"perception difficulty level"

AT&T video fails at Social Media

attWhile many companies seem to be understanding how social media works and they are slowly getting better at tracking consumer sentiment and complaints they still seem to fail at responding.  Following someone's bad advice as to how to best respond to the increasing negative feelings within social media and mainstream media, AT&T produced a "Im a PC" style video.The canned response from AT&T is a PR polished video combining flashy graphics, friendly images and some random guy named Seth the blogger talking about how everything is not really that bad as supported by the random graphs shown in the background.  The video also highlights that AT&T is to be credited for enabling the smartphone revolution over the past few years with no supportive facts or recognition of the other network providers.The first point which Seth the blogger helps explain is that the issues around their current network issues are in fact based on you and 300% the growth in wireless usage each year.  Considering falling revenue from other segments of the market would the best way to explain your failing network to provide quality service on customers using it? Is it better than consumers are not using your network?Each time you use data on your mobile phone plan over your allocated data rate you pay your phone company its a simple enough concept.  Looking on a AT&T business rates plans this shows that excess data usage is charged at $0.0048/kb or $5.12/mb which makes for a very profitable business, not something to complain about.The process where AT&T begin to try and repair their reputation involves first explaining the complexities behind the issue of bandwidth and how more complex it is to deal with MMS.  This is another area where they fail down as typically the early adopters who are also the ones making the most complaints have a good grasp around the technology and don't need a "Dummies guide to mobiles" video. If you are trying to reach your audience using social media you need to tailor your message to your audience, and this is another area where AT&T fail.The point about this ground breaking technology is that MMS is not a new feature on mobile phones and has been available on GSM/GPRS since MMS was first introduced by Telenor of Norway in March 2002.  This was followed by Optus introduced MMS into Australia in July 2002, and mm02 launch in europe in October 2002, so why the big issue for a large telecommunication company like AT&T?As Gizmodo points out the failure of the MMS is likely due to AT&T Opt-Out codes automatically enabled on all their subscribers accounts, so if this is in fact an internal issue why build a smoke screen around the issue that they need to do more?So the points to learn from this social media failure is if your are responding to your complaints
  • match your message to your audience
  • don't include useless images/graphics
  • support your statements with facts
  • don't treat your audience as idiots
  • use high profile executives not characters such as "Seth the blogger"
  • don't spend more money on the video than your customer service
  • update your website to include new product information you refer to "MMS"

YouTube Video Ads

youtube logo

On of the biggest issues around video has been its been a typically one way medium, while people can comment back by text or video can it be used for viewer feedback.  Recently YouTube has shown the potential to offer multiple response options for advertisers interactive videos using the overlays linked to other videos. While it is not clear if this is a beta test or just a very smart and inventive person who is pushing the boundaries of the YouTube platform.

The video player is built in Adobe Flash so technically anything you can do in a standard flash application can be overlaid on user videos, and this is becoming increasingly common with popular videos.  The multiple choice survey test shown below is running on Machinima.com's YouTube channel where, they are broadcasting game previews from Comic Con 2009. They are very clear with different choices with individual colours and even a text box at the top of the screen to explain what they want you the viewer to do. 

youtube-rate the game

The "rate this game" option is shown towards the end of the game preview.  When one of the options are selected a new YouTube video loads, with a clear title as to which option you selected, but could the video instructions be a bit more clear for the next steps as you can see in the screenshot below. 

The reason I feel that the steps are not clear is that around 15-20% of people used the 5 star rating system to provide feedback and not the text comment box below the video. Although the text is explaining the call to action is very large, since this is a visual element, a short video showing the desired action would make sense but only because its YouTube.

If you were running this on different medium/platform such as your website screenshots or flash animation could deliver the same visual explanation as not all users maybe used to using the comment field.


So how do you measure the results, the video requests for comments to be made in the text comment field below, but this also requires the user to have an account and login.  Based on the chart below based on the YouTube comments reaction more people are unsure if they want the game so they must analyze the comments for more understanding.

An easier way to measure survey results is to track which of the three videos options get the most views, the benefit is that the viewer doesn't have to login to track their response.  From the chart below we can see that around 62% of views of the 3 options were that people want the game which may not be the perfect campaign result but to scale for large campaigns with thousands of comments its more efficient.

Comments can still be useful as they can provide feedback such as the "game lighting is too dark" or the "there appears to be glitches in background images" but the number of comments were a fraction of video viewings.  There is also the issue which is visible of users responding and interacting amongst themselves off the topic and even starting abusive attacks against other users commenting.

There are a few issues, one is that YouTube makes all videos public, so once one option such as "hate it" is selected it is easy to see related videos such as "not sure" and "want it". If campaigns like this are to be more transparent, it should not be possible that users can select one option and then click to view other options skewing your results.

The obvious issue for analysis based on the chart below is that the number of comments were fairly consistent between each option so do not provide as much feedback compared to the video viewings, the lower the number the higher influence of duplicate posts, and multiple posters, spam...


So how can web analytics help this test deliver better results for business?

Using methods such as A/B testing, they can test a different second trailer for the same game, using the same question but with 3 new video response pages.  This way they can test which particular scenes or clips generate the most positive reaction to the game.  This information can be further analysed as why but can be used to make a simple decision as to which would be best in an advertising campaign.

Further tests could be would you buy this game based on this preview, using the same call to action tracking and comparison of data to measure if a second trailer generated higher "want" results but also higher "buy me now" results.

Click to see the video Iron Man 2 Comin Con "Rate This Game"

Top YouTube Myths

youtube logoGoogle's YouTube seeks to counter the coverage is that it is just a source of viral videos, attention seekers and animal videos in a recent blog post.  YouTube is continuing to stress its recent agreements with content partners such as Sony, Diney & Universal music has raised its profile but what are the most common myths.Myth 1: YouTube is limited to short-form user-generated content. YouTube directors & media partners have uploaded hundreds of full-length feature films and TV Shows.Myth 2: YouTube videos are grainy and of poor quality. YouTube HD is now standard for hundreds of thousands of HD videos added each month.Myth 3: Traffic, growth, and uploads are bad for YouTube's bottom line. Google builds scaleable infrastructure so it depends on growth and welcomes the challenge.Myth 4: Advertisers are afraid of YouTube. 20th Century Fox took over YouTube's masterhead ads across YouTube homepages in 11 countries to push its recent X-Men movie, this Youtube campaign in the UK generated 15% uplift in brand awareness for X-Men.Myth 5: YouTube is only monetizing 3-5% of the site. Who made up this statistic?? YouTube's business model is focused on total number of monetised views, not % of site.YouTube media partners Mythbusters claim that it looks like these myths are busted for now, but YouTube will likely be requested to revisit these myths in the future.

Youtube Dominos Gross-out

This update is related to recent blog posts as it deals with viral video and how they sometimes cannot be beneficial to a companies image.In the last 24 hours a new viral video has started its rounds, it shows 2 Domino's staff involved in a Gross-out. The interesting thing is that the company was made away of the video when the AdAge article went to press, so based on the response to #amazonfail how would they react?Dominos decided not to issue a press release or post a statement online, it was felt that the company could deal with tens of thousands of impressions and a strong response from Domino's would alert more consumers to their video and cause more embarrassment.YouTube views7am - 20,000 9am - 51,000 2pm - 151,000 11pm - 690,000So is this still the correct response, when will companies acknowledge that once a video/story goes viral nothing can stop it quick enough.*Update* after 3 months people since this story was first covered, this blog post still attracts a decent number of visitors looking into find out about this issue.  What makes this important for understanding is that just because the viral campaign has stopped this ongoing article traffic shows that the potential fallout may have only just begun.
1 2