Snooping on YouTube

 

If you are curious about something but nobody is willing to tell you, what do you do?

Recently, BBC reported on an interesting method of gathering market intelligence. The question researchers at the University of Massachussetts were trying answer was - how big is YouTube?

This may sound like a simple question but for various reasons - commercial, or otherwise - this is not the kind of information YouTube, or its parent company, Google, would like to place in the public domain. As the article puts it - 

In its 20 years of operation, YouTube has shaped entire generations' sensibilities and redefined global culture. Surveys show YouTube is the most popular social media site in the US by far, with 83% of adults and 93% of teenagers among its patrons. It's the second-most-visited website on Earth by most estimates, topped only by Google.com itself. But as the platform enters its third decade, the most basic facts about YouTube are still a closely guarded secret.

So how did the researchers at UoM get this information?

They started with the URLs that appear for almost all YT videos. They all start with  "youtube.com/watch?v=", and a unique 11-character identifier. Eg - Gangnam Style's identifier is 9bZkp7q19f0. 

This is where the raw computing power comes in. 11 random alpha-numeric characters can be arranged in more than 18.6 quintillion ways! So the researchers then wrote a program that would trawl the internet with random addresses and download the videos that scored a hit. It seems to collect the first 10,016 videos, the scraper tried more than 18 trillion potential URLs!

The conclusion?

When Google first acquired the platform in 2006, around 65,000 videos were uploaded every day. By 2022, they calculated that YouTube housed more than nine billion videos. By mid 2024, that number had grown to 14.8 billion videos, a 60% jump!!

In the cat and mouse game of corporates and snoops, perhaps it is just a matter of time before YT/Google comes up with ways to foil this kind data-collection and verification. 

--------------------------------------------

REFERENCES & LINKS

https://www.bbc.com/future/article/20250213-youtube-at-20-a-computer-that-drunk-dials-online-videos-reveals-statistics-that-google-doesnt-want-you-to-know

Lollipop or Marshmallow?

Earlier this week, I received a notification on my android phone. Google was offering to upgrade my mobile OS from Lollipop to Android 6.0 Marshmallow.

"Upgrade available!" is not exactly a great pick-up line - especially for those who have survived the whims and fancies of Microsoft OS upgrades. The message from Google also assumes that its customers are eagerly waiting for an upgrade. I am certainly not one of them. I am quite happy with my lollipop, and especially so when the advantages of moving from one confectionery to another is not quite clear.

Still, curiosity got the better of me and I checked the web for possible advantages of an upgrade. According to the Android homepage, Marshmallow offers -

• more control on permissions for apps
• fingerprint recognition - more secure
• more intuitive user interface
• better battery management 

Well, so?

I already have some control of permissions for apps from the Setup; Fingerprint recognition may help me in keeping my kids off my mobile, but do I really need it all the time? No. Also,"more" and "better" does not give me sufficient data on either the new user-interface, or battery management.

The final clincher comes from a number which is readily available: the download is 1657 MB. Data download costs money and when the download is more than 1.6 GB you certainly want to wait until there is spare data

Google also promises you that the upgrade process will take "a long time" during which the device will restart several times.

So there you are. Time and money. The Lollipop tastes even better now...

POSTSCRIPT (22 April 2016)

A few days ago, at the fag end of my monthly data subscription (25 GB), I decided to take the plunge and upgrade. Despite a connection speed of 5mpbs+ the download (1.78 GB) and upgrade took about eight hours.

Since I had been warned that all the contents of my SD card would be erased, I had taken a backup. Even after taking the required precautions, it is quite disappointing to see the following bugs after the upgrade from Lollipop to Marshmallow:

• CamCard - My entire collection of visiting-card scans has disappeared. I get an image of the cards on Google Contacts, but the CamCard app seems unprepared for Marshmallow
• Scheduled Power On & Off: After the upgrade my phone does not turn itself on at the designated time - 05:30 . Now it has to be turned on manually.
• Many app now show that they were "designed for an earlier version of Android"
• The contact list has got messed up - each entry seems to have got duplicated :((
• There is no sign of the promised "Fingerprint Security" option
• In "Scheduled Power On & Off", I have my mobile turned off between 12:00AM and 05:30AM. Ever since I upgraded to Marshmallow, the phone does not turn on automatically. I need to do this manually, every morning :(
• Everytime the mobile reboots, "Data Comm." gets turned on automatically, even if the WiFi is already on. Very annoying.

-----------------------------------------------------
LINKS

- https://en.wikipedia.org/wiki/Android_Marshmallow

- https://www.android.com/intl/en_in/versions/marshmallow-6-0/

In a Wynk

In our everyday lives, there are few things as magical as wireless networks.

Each time I swipe my credit card I find myself counting the seconds before the mobile phone in my pocket gives me a buzz. A few seconds for the transaction to go from the swiping device, over the city's telecom networks, across a satellite hovering 36,000 km away in the geostationary orbit, to check with a database sitting in a server on the other side of the world, before a digital confirmation finds its way back into my mobile phone -- all within about 10 seconds!

The magic works on much smaller scales as well. Bluetooth devices on speakers and earphones, IR remote controls for potato couches, and RFID tags for shoppers, workshops and marathon runners.

Last weekend I came across something new - Lenovo's app for high-speed, peer-to-peer sharing of files and folders: Shareit. I was walking out of the ADHM-2015 bib-distribution area when I saw a kiosk distributing free earphones. It was music company called Wynk and you had to download their app to get the earphones. The only hitch was that I had run out of my data quota for the month and I could not connect to the Google Playstore. "No problem", said the Wynk guy, "I can just share it from my mobile."

At first I thought this was another form of Bluetooth - near field communication and data transmission at 2.4 GHz. But no, this was not Bluetooth. Even with the WiFi and Bluetooth turned off, a file bigger than 15 MB jumped across thin air in less than three seconds! How did this happen?

Turns out that this 5 GHz band is not special to either Lenovo or Wynk. It is a part of the WiFi bandwidth that is less congested compared to the ones used by Bluetooth devices. So it is a lot faster but this transmission-frequency is efficient only at less than half the range of 2.4 GHz devices, which is less than 5 ft..

-------------------------------------------------
LINKS

* List of 2.4 GHz Radio Use -- https://en.wikipedia.org/wiki/List_of_2.4_GHz_radio_use

* What is the difference between a 2.4 GHz and 5 GHz WiFi? -- http://www.howtogeek.com/222249/whats-the-difference-between-2.4-ghz-and-5-ghz-wi-fi-and-which-should-you-use/

* Lenovo Shareit FAQs - http://shareit.lenovo.com/faqs.html

Google's 'Backrub'

One of the earliest search engines, Alta Vista, was given euthanasia this week. It had been terminally ill for so long that a tweeter said, "...it was like getting an invitation to a funeral of someone you thought died 10 years ago."

Alta Vista, by indexing around 20 million web pages  was the forerunner to the billions of pages that Google now indexes. Aparantly, its text-analysis approach just couldn't cope with spamming, and was not match for Google "backrub approach".

Now, what is this "backrub approach"??

BackRub was a research project of the Digital Library Project in the Computer Science Department at Stanford University. It was started by Larry Page under his supervisor Terry Winograd. The codes were written in Java and Python and ran on several Sun Ultras and Intel Pentiums running Linux, with its primary databased fitting into a 28GB disk.

Essentially what Page's program did was to find out which web pages link to a given page, considering the number and nature of such backlinks to be valuable information about that page (with the role of citations in academic publishing in mind). From backlinks came the title BackRub which later evolved into the now famous Google search engine.

---------------------------------------
LINKS

Khanna, Sundeep (2013): HOW ALTA VISTA LOST ITS MOJO AND GOOGLE FOUND ITS GROOVE, The Mint-WST, 5Jul13 -- http://www.livemint.com/Opinion/h5h8GJT1BuFgQY2qm0bxzI/How-AltaVista-lost-its-mojo-and-Google-found-its.html

ANATOMY OF A LARG-SCALE HYPER TEXTUAL SEARCH ENGINE -- http://infolab.stanford.edu/~backrub/google.html

BackRub -- http://web.archive.org/web/19971210065425/backrub.stanford.edu/backrub.html

BackRub FAQ -- http://web.archive.org/web/19971210065437/http://backrub.stanford.edu/FAQ.html

Maria's Insects & Flowers

 

Thanks to today's unusual doodle, here is the remarkable story of a Dutch lady who loved to draw insects and flowers:

The Guardian - GrrlScientist (2 April 2013) -- http://www.guardian.co.uk/science/grrlscientist/2013/apr/02/maria-sibylla-merian-artist-insects-science