Meme Mining for Fun and Profit

1
Meme Mining for Fun and Profit

• By Broward Horne
• DefCon 13

2
Geek Story

• My Flight From New Jersey

3
Me

• Why should you listen to me?
• Hacked some systems prior to 1991

4
Our Goal Ride the Technology Wave

• Operational S-curve theory
• Tactical MemeMiner Parlor Tricks
• Strategic Social Bandwidth

5
Dot Com Crash Continues?

• Recent layoffs in 2005
• IBM 13,000 Hypercom
• Sun another 5 BMC
• GM 25,000 Krispy Kreme
• GE reorganization Cray
• Alcoa 6,500 DHL
• Winn Dixie 22,000 American Express
• Lear 7,700

6
What To Do?

• Your time and effort are finite.
• The IT market changes constantly.
• What skills should you focus on?
• A mistake can be costly.

7
Technology Meme Propagation

• Exponential Function
• 1 -gt 2 -gt 4 -gt 8, etc
• Until population saturation
• Known generically as an S-Curve

8
S-Curve

• Fundamental business function

9
Detailed S-Curve

• Inflection Point is 50 mark

10
S-Curve Predictive Power

• Change in Rate of Change is key

11
Peak Oil

• Audience What is Peak Oil?
• Hubbert used S-curve to predict Peak Oil event in
  United States

12
Harry Dent

• Great Boom Ahead S-Curve

13
Spawned S-Curves

• Newbury Networks Wifi Tracking System

14
Meme Mining Theory

• Frequency count is proxy for S-curve
• Monitor technical trends
• Poor Mans Data Miner Equalizer

15
Meme Miner

• Simple java program (show code)
• Executes sequential queries
• Builds chart with query results

16
Historical Graphs

• Work can you identify the pattern?

17
Historical Graphs

• Easter Bunny

18
Historical Graphs

• Comdex shows no recovery

19
Historical Graphs

• High correlation to stock price

20
Historical Graphs

• Palladium rumors precede rise (Latency)

21
Historical Graphs

• Shrinking Latency?

22
Historical Graphs

• Classic S-curve points of interest

23
Real-Life Usage

• Delphi Programming Language

24
Pre-inflection Graphs

• Trends in Scripting Languages

25
Pre-inflection Graphs

• WIMAX about to explode?

26
Pre-inflection Graphs

• Blogs ( 27 dont know what it is )

27
Pre-inflection Graphs

• Wi-fi phone is hot

28
Pre-inflection Graphs

• Inflection tends to be at 50 point
• Inflection point is maximum hysteria
• What can we predict about blogging?

29
Pre-inflection Graphs

• RSS Feeds

30
Miner Deficiencies

• Requires knowledge of trend to test
• Subject to manipulation
• Not statistically rigorous
• Dependent on good keywords
• Works poorly for non-technical trends

31
Potential Improvements

• Use RSS feeds to catch trend earlier
• Skewing factor for search engines
• Extract more abstract relationships

32
Moores Law

• Justification for continued IT boom
• But are there other limiting factors?

33
Social Bandwidth

• People have finite bandwidth
• People have finite storage
• Shared context impacts bandwidth

34
People Spend Time On

• Activities they enjoy
• Activities theyre forced to do
• All others are discarded over time

35
Average Person
36
Paris Hilton
37
Internet S-Curve

• Has Internet Growth peaked?
• What happens if growth peaks?

38
DejaNews Frequency Count

• Rough Proxy for Internet Growth

39
Major Events
40
DejaNews Anomaly

• Market saturation?

41
Dejanews Falloff

• Shift to blogging
• Changes in Google methodology
• Slowing rate of change in Net Growth

42
Radical Concepts?

• POI Price Of Information
• COI Cost Of Information
• Schramms Model Context
• nate_at_metricaldesign.com

43
High Level Model

• The Information Technology Boom

44
Cost of Information

• Marginal Cost versus Marginal Return

45
Cost Versus Price

• Diversity versus Cost

46
IT Trend Change

• Marginal cost exceeds return

47
Schramm Model
48
Schramm Derivatives

• Reduce COI by reducing transactions
• Reduce COI through group context
• Microbrewery model (current employer)
• Restaurant Management (UNLV)