Presentation is loading. Please wait.

Presentation is loading. Please wait.

Measuring the Growth of Work As input size N increases, how well does our automated system work? –Depends on what you want to do! Use algorithmic complexity.

Similar presentations


Presentation on theme: "Measuring the Growth of Work As input size N increases, how well does our automated system work? –Depends on what you want to do! Use algorithmic complexity."— Presentation transcript:

1 Measuring the Growth of Work As input size N increases, how well does our automated system work? –Depends on what you want to do! Use algorithmic complexity theory: –Use measure big o: O(N) which means worst case Important for –Search engines –Databases –Social networks –Crime/terrorism Sub-linear O(Log N) Linear O(N) Nearly linear O(N Log N) Quadratic O(N 2 ) Exponential O(2 N ) O(N!) O(N N ) Performance classes Polynomial Death to scaling

2 How Much Information Reaches Americans Consumers? Roger Bohn - Rbohn@ucsd.eduRbohn@ucsd.edu HMI? Project -- Global Information Industry Center (GIIC) UC San Diego November 3, 2009

3 HMI? 2009 Scope Consumers - Information used by Americans 1960 to 2008 Enterprise - Information processed by servers worldwide in 2008 Special studies, such as Scientific data at MIT - giic.ucsd.edu Health-care data Enterprise storage growth

4 HMI? Web site papers

5 Defining “information” Dozens of definitions (more than one per person!) Information to consumers ≃ Data that is directly for people Information supplied versus consumed Whether they pay attention or not (multiple streams count fully) Multiple metrics for measuring an information object Bytes: emphasizes moving visual imagery Info C Words: ignores pictures Info W

6 How many bytes in a 20 minute TV show? Raw bytes = Screen size in pixels x bytes per pixel x frames per sec. 1920 x 1080 x 3 x 30 = 186 Megabytes per second = 1.5 Gbps TV compressed down to 1 to 2% e.g. 12 Mbps Info compressed counts “bytes through the cable” Examples: SDTV ~ 4 Mbps = 2 GB per hour Book = 12 bytes per word x 4 words per second = 400 bps

7 Methodology

8 How do we consume information? Source: HMI? 2009 Info c Info w

9 Information consumed by “average American” per day Info C in GB/dayInfo W in words/day TV (incl. DVR, Internet, mobile)12.044,342 Radio0.18,315 Phone0.015,269 Print0.018,659 Computer0.0827,122 Computer games18.52,459 Movies3.3198 Recorded music0.081,112 Total34.097,476

10 Information consumed by “average American” per day Info C in GB/day TV (incl. DVR, Internet, mobile)12.0 Radio0.1 Phone0.01 Print0.01 Computer0.08 Computer games18.5 Movies3.3 Recorded music0.08 Total34.0

11 Computer games ubiquitous 35 M consoles sold in 2008; 10M Nintendo DS + 10 M Wiil 70% of online adults play computer games 7 of top 10 iPhone apps are games (by revenue)! “Casual gaming” has removed gender bias

12 Sub-categories: computer games # of Users (millions) Average use/monthTotal Info H % of Total Info H % of Total Info C PC (high performance)2186.9 hours1,4051.71%38.28% PC (standard)12418.11942.12%5.29% Consoles8930.33682.55%10.02% Handheld Devices12912.6241.54%0.64% Total 19917.92%54.23% Full calculations go deeper: types of users, types of games, compression factors, others

13 How push so many bytes?

14 Detail, fluid animation, lighting, good AI. This is from Xbox version - PC graphics better

15 Slices: How is info delivered? Airwaves includes TV broadcast, radio, mobile; Wires includes Cable TV, Internet; Physical = Print + Local computer

16 Other use Slices “Interactive” information = telephone, computer games now 55% of Info C, 36% of Info W Traditional media = Cable TV, broadcast TV, radio, movies, print Now now 36% of Info C, 56% of Info W Define groupings of media types and information measures

17 What about Information Growth? As costs fall, use increases. Hypothesis: ~ 35% per year growth in Info C Incorrect!

18 slow Growth 1960 to 19801980 to 2008 INFO Hours 3.9%2.6% INFO Words 3.7% 3.0% INFO C (bytes)2.9%5.4% US Population1.1%1.0% GDP $/capita3.6%2.9% Source: HMI? 2009

19 Decomposing growth 1960-1980 Info C = (# Users) x (Average Info H per user) x (Average bps C )x 3600/8 ∆Info C = ∆Population + ∆(Info H per capita) + ∆bps C »Where ∆Y = (dY/dtime)/ Y = % rate of change of Y 5.4% = 1.0% + 1.6% + 2.8% = Population growth + Hours/person growth + Info C intensity growth Why did average bits per second only grow at 2.8% per year?

20 Why such slow growth? TV penetration ~97% in 1980 Hours per day: 7.4 in 1980 11.7 in 2008 Intensity: 2.9 Mbps in 1980 6.5 Mbps in 2008 TV in 1980: 4 Mbps SDTV TV in 2008: 4 Mbps SDTV 12 Mbps HDTV < 1 Mbps Internet

21 what’s happened to print media? Print media = Magazines + books + newspapers Print = fastest way to deliver words Pool: 240 words per minute 24% of I words in 1960; 9% in 2008 Web browsing another 28% Television 1960 1980 2008 Print Computer e.g. Web browse Telephone has grown in Share w Source: HMI? 2009 Info W distribution across media

22 Roll your own projections What happens if cable TV increases bandwidth per channel? TV usage migrates to Hulu? What if DVR users store 1% of what they watch, at high fidelity, for 2 weeks? Quality versus quantity trends

23 Rise of the Internet

24 Information Growth As costs fall, use increases. If ≅ 1, quantity will rise at same rate as price falls Cost per transistor

25 Variable bit rates between and within categories TV (cable, DVD, etc.)3,000 (DVR) to 11,000 (DVD) to Print.018 (color magazine) to.001 (straight text) Radio.019 to.096 Phone.064 (landline) to.010 (cellular) Computer non-game100 (average Internet broadband) Computer game23,000 (console) to 2000

26 Two Categories of Algorithms 2 4 8 16 32 64 128 256 512 1024 Size of Input (N) 10 35 10 30 10 25 10 20 10 15 trillion billion million 1000 100 10 N N5N5 2N2N N Unreasonable Don’t Care! Reasonable Runtime sec Lifetime of the universe 10 10 years = 10 17 sec


Download ppt "Measuring the Growth of Work As input size N increases, how well does our automated system work? –Depends on what you want to do! Use algorithmic complexity."

Similar presentations


Ads by Google