程序代写代做代考 chain DNA AI distributed system flex assembly graph concurrency game GPU html algorithm Future of Computing and

Future of Computing and
What Do We Do When We Get There?
April 30, 2020
15213 s’20
© 2016-20 Goldstein 1

Today
• Beyond Moore’s Law • Technology & Labor • What is Money?
• 213 Final
15213 s’20 © 2016-20 Goldstein 2

Moore’s Law Origins
1965: 50 1970: 1000
• Moore’s Thesis
– Minimize price per device
– Optimum number of devices / chip increasing 2x / year
• Later
– 2x / 2 years
– “Moore’s Prediction”
15213 s’20
© 2016-20 Goldstein 3

1.E+10 1.E+09 1.E+08 1.E+07 1.E+06 1.E+05 1.E+04
1.E+03 1970
1980 1990
Year
2000
2010
Desktop
Embedded
GPU
Server
General Trend
Moore’s Prediction
Sample of
117 processor chips
Moore’s Law: 50 Years
Transistor Count by Year
15213 s’20
© 2016-20 Goldstein
4
Transistors

Moore’s Law
Time
15213 s’20 © 2016-20 Goldstein 5
Goodness

Moore’s Law
Time
15213 s’20 © 2016-20 Goldstein 6
Goodness

Moore’s Law
Time
15213 s’20 © 2016-20 Goldstein 7
Goodness

Moore’s Law
Happy Happy
B
B‘
’day D
ay
Time
15213 s’20 © 2016-20 Goldstein 8
Goodness

Eniac → PS/4
• How much would enough Eniac’s weigh
to equal 2.8Kg of PS/4 computing?
=?
15213 s’20 © 2016-20 Goldstein 9

Eniac → PS/4
• How much would enough Eniac’s weigh
to equal 2.8Kg of PS/4 computing?
A little perspective. Empire State Building weighs ~2.7×108 Kg
? Alternatively, more than all
the buildings in Pittsburgh!
1000 Empire State Buildings of Eniac’s!!!
15213 s’20 © 2016-20 Goldstein 10

Eniac → PS/4 • From 1946 to 2014
Ops/sec
Eniac
5×103
PS/4
2×1010
Improvement
106
Cost 6×106 $ 4×102 $ 104 An improvement of
Power 1.5×105 W 1×102 W 103 1024 ops/sec-$-Kg-m3-W
Volume 6.5×102 m3 4.5×10-3 m3 105
Weight
2.7×105 Kg
2.8 Kg
105
15213 s’20 © 2016-20 Goldstein 11

Understanding Exponentials
• Key to future forcasting
• Very Very hard for humans to do
15213 s’20 © 2016-20 Goldstein 12

Understanding Exponentials
• Key to future forcasting
• Very Very hard for humans to do
• Example: Kasparov Vs. Deep Blue
– 1989: Kasparov destroys deep blue
15213 s’20 © 2016-20 Goldstein 13

Understanding Exponentials
• Key to future forcasting
• Very Very hard for humans to do
• Example: Kasparov Vs. Deep Blue
– 1989: Kasparov destroys deep blue – 1996: Deep Blue wins one game
15213 s’20 © 2016-20 Goldstein 14

Understanding Exponentials
• Key to future forcasting
• Very Very hard for humans to do
• Example: Kasparov Vs. Deep Blue
– 1989: Kasparov destroys deep blue – 1996: Deep Blue wins one game
– 1997: Deep Blue wins tournament
15213 s’20 © 2016-20 Goldstein 15

Understanding Exponentials
• Key to future forcasting
• Very Very hard for humans to do
• Example: Kasparov Vs. Deep Blue
– 1989: Kasparov destroys deep blue – 1996: Deep Blue wins one game
– 1997: Deep Blue wins tournament
• Particularly hard in the beginning
15213 s’20 © 2016-20 Goldstein 16

What Moore’s Law Has Meant
• 1976Cray1
– 250 M Ops/second
– ~170,000 chips
– 0.5B transistors
– 5,000 kg, 115 KW
– $9M
– 80 manufactured
• 2017iPhoneX
– >10BOps/second
– 16chips
+ GPS
+ Microphone
+ Camera
– 4.3Btransistors(CPUonly)
– 174g,<5W – $999 + Wireless + graphics 15213 s'20 © 2016-20 Goldstein 17 – ~3millionsoldinfirst3days ++++++ What Moore’s Law Has Meant • 1965 Consumer Product • 2017 Consumer Product Apple A11 Processor 4.3B transistors 15213 s'20 © 2016-20 Goldstein 18 What Moore’s Law Could Mean • 2017 Consumer Product • 2065 Consumer Product – Portable – Lowpower – Willdrivemarkets& innovation 15213 s'20 © 2016-20 Goldstein 19 Using All Those Transistors 15213 s'20 © 2016-20 Goldstein 20 Insatiable Demand for Computing • Programmable matter? • Simulating life • Augmented Reality • Many many many more 15213 s'20 © 2016-20 Goldstein 21 Requirements for Future Technology • Must be suitable for portable, low-power operation – Consumer products – Internet of Things components – Not cryogenic, not quantum • Must be inexpensive to manufacture – Comparable to current semiconductor technology • O(1) cost to make chip with O(N) devices • Need not be based on transistors – Memristors, carbon nanotubes, DNA transcription, ... – Possibly new models of computation – But, still want lots of devices in an integrated system 15213 s'20 © 2016-20 Goldstein 22 Moore’s Law: 100 Years Device Count by Year 1.E+17 1.E+15 1.E+13 1.E+11 1.E+09 1.E+07 1.E+05 1.E+03 1017 devices! Desktop Embedded GPU Server General Trend Moore's Prediction 1970 1980 1990 2000 2010 2020 2030 2040 2050 2060 Year 15213 s'20 © 2016-20 Goldstein 23 Transistors Chips Have Gotten Bigger Intel 4004 1970 2,300 transistors 12 mm2 Apple A11 2017 4.3 B transistors 89 mm2 NVIDIA GV100 Volta 2017 21.1 B transistors 815 mm2 15213 s'20 © 2016-20 Goldstein 24 Chip Size Extrapolation Area by Year 16384 8192 4096 2048 1024 512 256 128 64 32 16 8 4 1970 1990 2010 147 cm2 15213 s'20 © 2016-20 Goldstein 25 Year 2030 Desktop Embedded GPU Server Trend 2x every 9.5 years 2050 Area (mm^2) Extrapolation: The iPhone XXX Apple A59 2065 1017 transistors 147 cm2 15213 s'20 © 2016-20 Goldstein 26 Transistors Have Gotten Smaller – Area A – N devices – Linear Scale L L 15213 s'20 © 2016-20 Goldstein 27 Linear Scaling Trend 1,000,000 100,000 10,000 1,000 100 10 1 1970 Submillimeter Linear Spacing by Year Desktop Embedded GPU Server Trend Submicrometer 1975 1980 1985 1990 1995 2000 Year 2005 2010 2015 2020 15213 s'20 © 2016-20 Goldstein 28 Sqrt(A/N) (nm) Submillimeter Dimensions 10-3 10-4 10-5 10-6 1 millimeter (mm) 500μm: 72μm: 50μm: 10μm: 5μm: 2μm: Length of amoeba Intel 4004 linear scale Average size of cell in human body Thickness of sheet of plastic food wrap Spider silk thickness E coli bacterium length 1 micrometer (μm) 15213 s'20 © 2016-20 Goldstein 29 Submicrometer Dimensions 10-6 1 micrometer (μm) 10-7 10-8 10-9 400-700nm: Visible light wavelengths 1 nanometer (nm) 211nm: 144nm: 30nm: 9nm: 2nm: 1nm: Apple A8 linear scale Apple A11 linear scale Minimum cooking oil smoke particle diameter Cell membrane thickness DNA helix diameter Carbon nanotube diameter 15213 s'20 © 2016-20 Goldstein 30 Linear Scaling Extrapolation 100,000.0 10,000.0 1,000.0 100.0 10.0 1.0 0.1 1970 Desktop Embedded GPU Server Trend Linear Scale by Year 1990 2010 2030 Year 2050 230 pm 15213 s'20 © 2016-20 Goldstein 31 Sqrt(A/N) (nm) Subnanometer Dimensions 10-9 10-10 10-11 1 nanometer 1nm: Carbon nanotube diameter Silicon crystal lattice spacing 2065 linear scale projection Spacing between atoms in hydrogen molecule Electron-proton spacing in hydrogen (Bohr radius) (nm) 543pm: 230pm: 74pm: 53pm: 2.4pm: Electron wavelength (Compton wavelength) 1 picometer 10-12 (pm) 15213 s'20 © 2016-20 Goldstein 32 Pause for a second • 50 more years of Moore’s law? – Probably not in Si, but ... • Probably need new architectures, etc. • But, so far, Necessity is the mother of invention 15213 s'20 © 2016-20 Goldstein 33 1.E+11 1.E+10 1.E+09 1.E+06 1.E+03 1.E+00 1.E-03 What Comes Next? Ops/sec/$ doubles every few months? nanometer Manufacturing plays a key role! doubles every 1.0 years Tubes/ Transistor CMOS doubles every 2.3 years 1.E-06 1880 Mechanical/ Relays doubles every 7.5 years 1900 1920 Combination of Hans Moravac + Larry Roberts + Gordon Bell WordSize*ops/s/sysprice 1940 1960 1980 2000 2010 2020 2030 15213 s'20 From Gray Turing Award Lecture © 2016-20 Goldstein 34 What This Means for You • As Computer users and Scientists? • As People? 15213 s'20 © 2016-20 Goldstein 35 Great Reality • More Heterogeneity – Centers of Racks of Systems of Boards of Chips of • Cores • GPUs • FPGAs • Lots of Concurrency 15213 s'20 © 2016-20 Goldstein 36 Exponentials • Moore’s Law • Biology • Big Data/AI • Manufacturing – Computing – Synthetic Biology – Additive Manufacturing • Robotics • Energy 15213 s'20 © 2016-20 Goldstein 37 Exponentials Abound Human 2.0 “Moore’s law” Sequencing Synthesizing $0.01/base, halving each year 106 bases, 10x per decade 15213 s'20 © 2016-20 Goldstein 38 Big Data can look like AI Big Data ML Interesting patterns 15213 s'20 © 2016-20 Goldstein 39 Big Data can look like AI Big Data ML More Interesting patterns 15213 s'20 © 2016-20 Goldstein 40 Big Data can look like AI Even Bigger Big Data ML Patterns that look like Creativity/Intelligence 15213 s'20 © 2016-20 Goldstein 41 What is you prediction? AI-ish? 1.E+13 From Forbes.com: Speech->text amazing Translation getting pretty good
1.E+12
1.E+11
an expected dip in profit, analysts are generally optimistic about Fluor as it prepares to
Despit
e
1 1.E+09
.E+10
Image classification
reports its first-quarter earnings on Thursday,
1.E+08
May 1, 2014. The consensus earnings per share
1.E+07
estimate is 96 cents per share.
1.E+06
1.E+05 1.E+04
1.E+03
route to a win, as it leads 51-10 after the third
q
1.E+00
“WISCONSIN appears to be in the driver’s seat en
1.E+02
1
a
u
rter. Wisconsin added to its lead when Russell Wilson fohutntdp:/J/acrcohbivPee.idcse.ruscei.nedfuo/rmla/ndaetaigshet-sy.hatrmdl
0 20 40 60 80 100 120 140
.E+
01
touchdown to make the score 44-3 … . ”
15213 s’20
© 2016-20 Goldstein
42
1987
2011

Exponentials
• Moore’s Law
• Biology
• Big Data/AI
• Manufacturing – Computing
– Synthetic Biology
– Additive Manufacturing
• Robotics
• Energy
15213 s’20
© 2016-20 Goldstein 43

The World is Changing
• Technology Trends
– Nanotechnology
Clean ubiquitous Cheap Energy
More Knowledge available to more people – Synthetic Biology
– Computers – Robotics
Safer & Faster travel
• Some Implications
• What Should We Do?
A level playing field
Customized anything Better Health Dematerialization of Value
15213 s’20 © 2016-20 Goldstein 44

• Technology Trends
– Nanotechnology
Clean ubiquitous Cheap Energy
Utopia?
More Knowledge available to more people – Synthetic Biology
– Computers – Robotics
Safer & Faster travel
• Some Implications
• What Should We Do?
A level playing field
Customized anything Better Health Dematerialization of Value
15213 s’20 © 2016-20 Goldstein 45

15213 s’20 © 2016-20 Goldstein 46

KiTllerchRnooblogtsy?& RJeoabllsy? It’s a Lot More Complicated.
15213 s’20 © 2016-20 Goldstein 47

Today
• Beyond Moore’s Law • Technology & Labor • What is Money?
• 213 Final
15213 s’20 © 2016-20 Goldstein 48

Technology & Jobs
• Replace Workers
15213 s’20 © 2016-20 Goldstein 49

Technology & Jobs
• Replace Workers • Eliminate Jobs
15213 s’20 © 2016-20 Goldstein 50

Technology & Jobs
• Replace Workers • Eliminate Jobs
• Enhance Workers
15213 s’20 © 2016-20 Goldstein 51

Technology & Jobs
• Replace Workers • Eliminate Jobs
• Enhance Workers • Create New Jobs
15213 s’20 © 2016-20 Goldstein 52

Some Questions
• Is this time different?
• Why no significant job loss?
• Why is rate of GDP growth decreasing? • What is net effect?
• If there is an effect, When?
15213 s’20 © 2016-20 Goldstein 53

Some Questions
• Is this time different?
• Why no significant job loss?
• Why is rate of GDP growth decreasing? • What is net effect?
• If there is an effect, When?
15213 s’20 © 2016-20 Goldstein 54

Technology Revolutions
• TR1: (aka First Industrial Revolution) – Coal powered steam engine
– Cotton Gin
– Steam-powered printing presses
1760-1830
1965-?
15213 s’20 © 2016-20 Goldstein 55

Technology Revolutions
• TR1: (aka First Industrial Revolution) – Coal powered steam engine
– Cotton Gin
– Steam-powered printing presses
• TR2: (aka Second IR)
– Electricity/Internal Combustion Engine
– Assembly Lines
– Telegraph/Telephone
1760-1830
1860-1910
15213 s’20 © 2016-20 Goldstein 56

Technology Revolutions
• TR1: (aka First Industrial Revolution) – Coal powered steam engine
– Cotton Gin
– Steam-powered printing presses
• TR2: (aka Second IR)
– Electricity/Internal Combustion Engine
– Assembly Lines
– Telegraph/Telephone
• Third Technology Revolution – Renewables
– Manufacturing without assembly – Computation
1760-1830
1860-1910
1965-?
15213 s’20
© 2016-20 Goldstein 57

• • •
Just Like TR1 and TR2 TR3 will:
Improve productivity Increase wealth (on average) Cause massive disruption
15213 s’20
© 2016-20 Goldstein 58

But, TR3 is different
• Continued exponential improvement
(E.g., Computation, Networking, Renewable energy, Synthetic biology, ML&BD, Robotics, …)
“Moore’s law”
Sequencing
Synthesizing
$0.01/base, halving each year
106 bases, 10x per decade
15213 s’20
© 2016-20 Goldstein 59

But, TR3 is different
• Continued exponential improvement
(E.g., Computation, Networking, Renewable energy, Synthetic biology, ML&BD, Robotics, …)
Don’t get confused.
It is not just about computation.
Assembly-Free Manufacturing!
15213 s’20 © 2016-20 Goldstein 60

10000 1000 100 10 1 0.1
0.01 1980
15213 s’20
1990 2000
2010 2020 © 2016-20 Goldstein Year
2030
2040
Solar Power in US
61
Watts (109)

But, TR3 is different
• Continued exponential improvement
(E.g., Computation, Networking, Renewable energy, Synthetic biology, ML&BD, Robotics, …)
• Technology is able to do “uniquely human” cognitive and physical tasks.
15213 s’20 © 2016-20 Goldstein 63

But, TR3 is different
• Continued exponential improvement
(E.g., Computation, Networking, Renewable energy, Synthetic biology, ML&BD, Robotics, …)
• Technology is able to do “uniquely human” cognitive and physical tasks.
• Labor component of the marginal cost of production  0!
15213 s’20 © 2016-20 Goldstein 64

But, TR3 is different
• Continued exponential improvement
(E.g., Computation, Networking, Renewable energy, Synthetic biology, ML&BD, Robotics, …)
• Technology is able to do “uniquely human” cognitive and physical tasks.
• Labor component of the marginal cost
of production  0!
(E.g., how much labor to produce and deliver the next
copy of an eBook?)
15213 s’20
© 2016-20 Goldstein 70

• 15213 s’20

• •
But, TR3 is different
Continued exponential improvement
(E.g., Computation, Networking, Renewable energy, Synthetic biology, ML&BD, Robotics, …)
Technology is able to do “uniquely human” cognitive and physical tasks.
Labor component of the marginal cost
of production  0!
(E.g., how much labor to produce and deliver the next
copy of an eBook?)
Unlike TR1 and TR2, TR3 will create
massive technological unemployment
© 2016-20 Goldstein 71

Not Much Work Now for Horses
TR2: Foreverendedthehorseasa factor of production.
15213 s’20 © 2016-20 Goldstein 72

Not Much Work Now for Horses
15213 s’20 © 2016-20 Goldstein 73
With respect to human labor:
Is TR3 the last Technology Revolution?

Wait a Second!
• Where are the exponential job losses?
15213 s’20 © 2016-20 Goldstein 74

Losses yes, Exponential no
No real change in median income for 40 years!
15213 s’20 © 2016-20 Goldstein 75

Wait a Second!
• Where are the exponential job losses? • What about new jobs?
15213 s’20 © 2016-20 Goldstein 76

What Are The New Jobs?
• Yes, there will be traditional jobs which require, e.g.,
– Creativity
– Empathy
– Entrepreneurship – Social Skills
Aside: Requires changing our educational goals and methods!
15213 s’20
© 2016-20 Goldstein
77

What About New Jobs?
• Yes, there will be traditional jobs which require, e.g.,
– Creativity
– Empathy
– Entrepreneurship – Social Skills
• But, as technology improves it will continue to take away jobs
Department of Artificial Empathy? 15213 s’20 © 2016-20 Goldstein 78

What About New Jobs?
• Yes, there will be traditional jobs which require, e.g.,
– Creativity
– Empathy
– Entrepreneurship – Social Skills
• But, as technology improves it will continue to take away jobs
• And, those that remain are low paying. 15213 s’20 © 2016-20 Goldstein 79

Wait a Second. Hmmm.
• Where are the exponential job losses?
Around the corner?
• What about new jobs?
How long do they remain in human domain?
• Rate of productivity increase has slowed
Not, really.
Average doesn’t tell the whole story.
15213 s’20 © 2016-20 Goldstein 80

Remember the Horses …
With respect to human labor:
TR3 is the last Technology Revolution
15213 s’20 © 2016-20 Goldstein 81

Plenty of Resources
• TR3 economy produces an abundance of goods and services
• Most people are unemployable
– Most jobs are substituted or destroyed – Those that remain are low pay
– (some are created for a limited time.)
15213 s’20 © 2016-20 Goldstein 82

What Should You Do Today?
• Study for 213 final.
15213 s’20 © 2016-20 Goldstein 83

What Should You Do Today?
• More seriously: – Learn to learn
– Be flexible
– Expect to continue to learn
• Its not only STEM
• In fact, soft skills dominate as most important components of high-end jobs.
15213 s’20
© 2016-20 Goldstein 84

What does this mean?
15213 s’20 © 2016-20 Goldstein 85

Horror or Opportunity?
• TR3 poses 2 existential questions for most everyone:
– How to get the resources needed for life? – How to find meaning and dignity?
We Need a Change in Perspective
15213 s’20 © 2016-20 Goldstein 86

Plenty of Resources
• TR3 economy produces an abundance of goods and services
• Most people are unemployable (Technology is just cheaper, better, … )
• What do they do?
15213 s’20 © 2016-20 Goldstein 87

What will we do?
• Activities which require either:
– Super-Creativity
– Empathy on the part of the consumer
• Play
YOU
How Will We Do it?
15213 s’20 © 2016-20 Goldstein 88

Today
• Beyond Moore’s Law • Technology & Labor • What is Money?
• 213 Final
15213 s’20 © 2016-20 Goldstein 89

A new look at Money
• What makes money work: – A metric of value
– A method of accounting
– Trust that it is transferable
15213 s’20 © 2016-20 Goldstein 90

Today’s Money
• What makes money work: – A metric of value
– A method of accounting
– Trust that it is transferable
• So today’s currency is: – A static token
– Backed by the state
15213 s’20 © 2016-20 Goldstein 91

Today’s Money
• What makes money work: – A metric of value
– A method of accounting
– Trust that it is transferable
• So today’s currency is: – A static token
– Backed by the state
But remember, it has no intrinsic value!
15213 s’20 © 2016-20 Goldstein 92

• 15213 s’20
• •

TR3 & Money
Currency doesn’t need to be static token. Nor backed only by the state.
Can we harness reputation with computation and networks to create a new form of money?
Goal: Monitize IOUs
(I.e., Micro-bearer bonds)
© 2016-20 Goldstein 93

Before TR3
• Irish Banking Crisis (‘60, ’70, ‘76) – Banks closed
– Checks can’t clear
– People used check as money
Good?
Reputation Authority
15213 s’20
© 2016-20 Goldstein 94

Before TR3
• Irish Banking Crisis (‘60, ’70, ‘76) – Banks closed
– Checks can’t clear
– People used check as money
• Banknotes
15213 s’20 © 2016-20 Goldstein 95

Before TR3
• Irish Banking Crisis (‘60, ’70, ‘76) – Banks closed
– Checks can’t clear
– People used check as money
• Banknotes
• Credit networks for resource allocation
15213 s’20 © 2016-20 Goldstein 96

TR3 & Money
• Currency doesn’t need to be static token. • Nor backed only by the state.
• Imagine that everyone can issue their own money backed by their reputation.
15213 s’20
© 2016-20 Goldstein
97
CMU Dollar
Seth Dollar

Who accepts Seth-Dollars?
• Certainly, I must.
• Anyone who expects to trade with me.
• People in my communities
• Most anyone who expects to trade with
someone I trade with, etc.
• The further the separation the less likely.
• Everyone will decide their risk tolerance.
Seth Dollar
15213 s’20 © 2016-20 Goldstein 98

Who accepts Seth-Dollars?
• Certainly, I must.
• Anyone who expects to trade with me.
• People in my communities
• Most anyone who expects to trade with
someone I trade with, etc.
• The further the separation the less likely.
• Everyone will decide their risk tolerance.
Currency is “executable” and can have means to reduce risk.
Seth Dollar
15213 s’20 © 2016-20 Goldstein 99

We Are All
Members of Multiple Communities
Network & ML replaces Publican
15213 s’20 © 2016-20 Goldstein 100

Money Requirements
• Transactions must be: – Secure
– Fast
– Low cost
– Concurrent (High BW for system)
• Currency itself must be – Forgery proof
– Maintain value
• Controllable by Central Bank
15213 s’20 © 2016-20 Goldstein 101

Money Requirements
• Transactions must be:
– Secure  – Fast  – Low cost  – Concurrent (High BW for system)
• Currency itself must be – Forgery proof
– Maintain value
• Controllable by Central Bank 15213 s’20 © 2016-20 Goldstein



102

Money Requirements
• Transactions must be:
–Secure   –Fast  –Lowcost   – Concurrent (High BW for system) 
• Currency itself must be
– Forgery proof   – Maintain value  
• Controllable by Central Bank   15213 s’20 © 2016-20 Goldstein 103

• • •

Blockchain
Used to track transactions Distributed
Simple rule to resolve data races maintain consistency – longest chain wins.
Implications:
– Need to wait to see if transaction went through
– Need to have copy of all blocks everywhere
15213 s’20
© 2016-20 Goldstein 104

Block chain
• Worldwide network has copies of the blockchain – a public ledger
• Longest chain is the official ledger
• For safety transactions must be
broadcast everywhere
15213 s’20 © 2016-20 Goldstein 105

Blockgraph: Principle of Locality
Pittsburgh
New York
Tokyo
Paris
Most transactions are local, so they stay local
15213 s’20 © 2016-20 Goldstein 106

Tied together with a global chain
Pittsburgh
New York
Global
Tokyo
Paris
15213 s’20
© 2016-20 Goldstein
107

Money Requirements
• Transactions must be:
–Secure   –Fast 
–Lowcost   Concurrent (High BW for system)  Currency itself must be
– Forgery proof   – Maintain value  
• Controllable by Central Bank   15213 s’20 © 2016-20 Goldstein 108

Shameless plug
• Interested in working on this for the summer?
• If so, Send me an email
15213 s’20 © 2016-20 Goldstein 109

Impacts of TR3
• Without Planning:
– Massive disruption
– Increased wealth inequality
– Bad stuff
• With Planning
– An amazing future
Seth Dollar
• Best chance for success is a bottom-up, distributed system that uses the market  Reinvent Money based on Reputation
15213 s’20 © 2016-20 Goldstein 110

Shameless Plug – part 2
• How to revive economy in time of pandemic?
• Businesses and People need loans to get by til everything re-opens
• Monitize crowd-sourced IOUs!
15213 s’20 © 2016-20 Goldstein 111

• • •
Shameless Plug – part 2
How to revive economy in time of pandemic?
Businesses and People need loans to get by til everything re-opens
Monitize crowd-sourced IOUs!
Seth Dollar
Interested? send me an email!
© 2016-20 Goldstein 112
• 15213 s’20

TAs for Fall 2020
• The instructors for 15/18-213/513/613
need you
please apply if you did well in 213 and want to be a TA
You?
https://www.ugrad.cs.cmu.edu/ta/F20/
15213 s’20 © 2016-20 Goldstein 113

Final Exams and Grade
• Please study well imitate machine learning
• Attend the Final Review
May 3, 6PM EDT, Zoom (see piazza for link)
• Remember the exam is cumulative
http://www.cs.cmu.edu/~213/exams.html
• Grading algorithm: (almost) no curving
exam/lab/quiz/PS distribution:
http://www.cs.cmu.edu/~213/syllabus/syllabus.pdf
weight of labs: http://www.cs.cmu.edu/~213/assignments.html
• Double-check your scores
get in touch with us if anything looks wrong in autolab
15213 s’20 © 2016-20 Goldstein 114

Course Evaluation
• Please fill out the course evaluation you should have and will receive email about it
• Please provide constructive comments this is your chance to make 213 better
• Please be fair with your scores
the university administration analyzes these numbers
• TA evaluation
https://www.ugrad.cs.cmu.edu/ta/S20/feedback/
15213 s’20 © 2016-20 Goldstein 115

Thank You!
15213 s’20 © 2016-20 Goldstein 116