Upload
olafur-andri-ragnarsson
View
136
Download
2
Embed Size (px)
Citation preview
Software
As computers became more powerful and more common, a new problem surfaced: software
Development of computers was a hardware problem
Software or programs did not get the same attention
Operating systems were primitive and programming was done at a very low level
“[The major cause of the software crisis is] that the machines have become several orders of magnitude more powerful!”
— Edsger Dijkstra, The Humble Programmer
Source:Software_crisis
Software Engineering was not a established field
Became known as The Software Crisis
The Software Crisis
IBM developed OS/360 for System 360
DEC developed VMS for VAX
Unix was grew out individual efforts as response to Multix
System V, BSD, Solaris
Minix was an academic effort, Linux grew out of frustration with Minix license
Operating Systems
FORTRANMathematical Formula Translation System Released in 1957
Higher level language that became breakthrough in writing software
Created by John Backus of IBM
Came on 2.000 punched cards Other languages followed: COBOL, Algol
Programming Languages
The Software IndustryFirst applications were non-serious
Soon business applications started to emerge
VisiCalc was the “killer-app” 20% of computer sales was due to this program
Other business apps appeared: Ledgers, payrolls, inventory, etc.
Disruptive technology
Killer AppsDan Bricklin and Bob Frankston Created VisiCalc, the first spreadsheet The spreadsheet created a new market
People bought the hardware to run the software
RPV
According to the RPV Theory, IBM would not be able to enter the PC market
Their customers were asking for big and powerful machines, and needed programs and support
Q2
IBM successfully entered the PC market – according to RPV theory this would be difficult. How did they do this?
IBM PC
IBM decided to enter the PC revolution
The company was loosing market share, competition was growing
Project “Chess”Bill Lowe was given one year to create a Personal Computer – “Acorn”
Lowe and his team – “Dirty Dozen”, went to work in Boca Raton, FL
Looked for parts outside of the company
The War of the OS
IBM needed an Operating System
Most popular system was Digital Research CP/M, created by Gary Kildall
Microsoft was providing programming languagesand suggested that IBM make a deal with DR
The War of the OS
IBM decided on PC-DOS from Microsoft which bought the OS from another company
Negotiated revenue sharing with IBM In the 80s, DOS had 90% of the OS market
IBM PC
The IBM PC was introduced 12. August 1981 in New York
4.7 MHz Intel 8088, 16 kb RAM
DOS 1.0
$1.565
Enter the Clones
IBM released all the specification of the machine Open system
This allowed new entrants to create IBM compatible machines Compac was one of them
Enter the ClonesIBM controlled the market for a few years
They rationalised their product lines - deliberately restricted performance of lower-priced models in order to prevent them from cannibalising higher-priced models
The Compac passed them in 1986 with the Intel 386 machines
The PC market took off
IBM started to loose market share
PC Compatible Machines Ruled
Early 80s IBM PC became the standard hardware
MS-DOS became the industry standard OS
Command Line Interface – CLI Text User Interfaces – TUI
The Demo in 1968Doug Engelbart at the Augmentation Research Centre in Melno Park
Demonstrated the future of computing
Features
A pointing device – the Mouse Hypertext, graphical user interface
Dynamic file linking
Shared-screen collaboration involving two persons at different sites communicating over a network with audio and video interface
Xerox Parc
Alto Computer 1972
Xerox created a lab in 1970
Palo Alto Research Park – PARC
PARC was a place for visionaries
The Alto computer system had Graphical User Interface – GUI and a mouse as an input
Desktop metaphor with Files and folders
Graphical User Interfaces – GUI
Steve Jobs visited Xerox PARC 1979 Negotiated at deal with Xerox
They showed him: Object Oriented ProgrammingComputer networksGraphical User Interface
Apple started to work on this vision The Pirate Years
Graphical User Interfaces – GUIDesktop metaphor
Point, Click, Drag
Files,foldersIcons
Windows,scrollbars
Menus
Graphicalfonts Clipboard,cutandpaste,undo
Point,activate,select
Apple LisaFirst commercial computer with a GUI
Introduced in January 1983 Cost $9.995
Motorola 68000 CPU at a 5 MHz clock rate and had 1MB RAM
Featured cooperative (non-preemptive) multi-tasking and virtual memory
Apple Lisa
First commercial computer with a GUI
Introduced in January 1983 Cost $9.995
Impact: Business failure Too expensive Too slow
MacintoshIn 1984, Apple launched Macintosh Cost $1.995
Graphical User Interface
This set the standard for Operating Systems
Specification: 128 KB of RAM Screen was a 9-inch, 512x342 pixel monochrome display
Macintosh
Acceptance was slow The Mac was underpowered The GUI required memory and power
Writing Software was difficult
Gained popularity in education and with graphical designers – desktop publishers
Not so popular in the traditional business sector Microsoft provided applications (office apps)
Others Join the GameMicrosoft launched Windows 1.01 in 1985
Gates and Microsoft believed Graphical User Interfaces were the future
Regarded Front-end to DOS
Other players IBM TopView, DR GEM
Impact Software companies ignored Windows The business sector was not ready
Windows 3.0Windows finally became usable Released May 1990
Better use of memory Multitasking Used the 286 and 386 hardware better Support for CD-ROM Solitaire
Impact: First GUI used by the PC market The end of DOS, finally
Windows 95
Microsoft turned to consumers Windows 95 was targeted at the consumer market Support for the Internet Internet Explorer Friendlier user interfaces
Impact
Released with great fanfare Came to dominate the OS market The OS become more important than the hardware
Lessons
▪ Shift from hardware to software▪ None of the minicomputer makers became a
significant factor in the desktop personal computer market
▪ The PC was disruptive technology▪ The minicomputer users were not buying PCs –
yet▪ This created a new set of entrants: Apple, Tandy,
Commodore, and IBM
▪ In the late 1980s the performance of PCs met the needs of minicomputer users
▪ This severely wounded minicomputer makers – many of them failed
▪ At same time IBM succeeded in entering the PC market – how?
▪ It created an autonomous organization in Florida – far away from it’s New York headquarters
▪ They created the PC market▪ Then headquarters took control and lost control to
the Clones
Lessons
▪ Xerox mangement did not enter the computer market
▪ PARC members tried to show management – but they “just didn’t get it”
▪ Xerox is in the copying documents business – their customers were not asking for computer systems
▪ Visionary Computers did not fit their resources, processes and values– RPV theory
Lessons
▪ Doug Englebart envisioned the future of computers
▪ Xerox PARC built the visionary computer – but did not pursue it
▪ Early enthusiast like Ed Roberts of MITS and others did not get rich of computers and software
▪ Visionaries like Dan Bricklin and Bob Frankston invented VisiCalc – did not make much money
Lessons
Lessons
▪ Bill Gates saw the potential of software and started Microsoft
▪ Took the opportunity with MITS▪ Focused on software▪ Gary Kildall invented the C/PM system but Microsoft
bought similar OS and succeeded▪ Wrote software for Apple and later Macintosh▪ You don’t have to have superior products to win▪ You don’t have to invent technology – just use it
Lessons
▪ Apple and Steve Jobs saw the potential of computers and then GUIs
▪ GUI were slow to appear▪ Infrastructure product - needs software and users▪ Stretched the hardware at the time▪ Disruptive with new market – consumers▪ Apple Lisa failed – lacking in performance▪ The Macintosh started slowly and found some niche
market in Desktop Publishing and schools
Lessons
▪ Windows 95 was marketed to the consumer▪ First mass market of Operating Systems– The Internet helped▪ Today we have three major Operating Systems– Linux (Unix based)– MacOS (Unix based)– Windows
1975 1980 1985 1990 1995 2000 2005
HardwareeraPC,Mac
SoftwareOSeraWindows,Office,MacOS
InternetHardwareConnects
IBMPC Microsoft
Apple
2010
SoftwareweberaWeb2.0,Social
2015
Internetofthings
PC Evolution
The Network is the Computer
The Internet cloud
More programs and data is stored on network servers
The Personal Computer becomes one of the form factors to access the network
Examples Amazon API Google Apps Facework Platform API
Eric Robot “The Gomshall Robot” 1928
• http://cyberneticzoo.com/robots/1928-eric-robot-capt-richards-english/
John McCarthy Marvin Minsky Claude Shannon Nathaniel Rochester
“…solve kinds of problems now reserved for humans…if a carefully selected group of scientists
work on it together for a summer”
It proved to be difficult to create truly intelligent software
Anything that worked was regarded as software, like search alorighms but not intelligence
For decades AI research went through springs and winters
Thus, ironically, AI has been very successful but at the same time failed
Symbolic AIThe earliest way to that researcher approached AI was manipulation of symbols
This is called "good old fashioned AI" or “GOFAI"
The theory was that human intelligence could be achieved by high-level symbolic or human-readable representations of problems
Many search algorithms grew out of symbolic AI
Out of this grew cognitive systems and expert systems
Expert Systems
Expert systems are systems that contain rules and facts
By answering series of questions users are lead to the conclusion based on the facts and the rules
Expert system require knowledge or data about a narrow specific field
Machine Learning
The general term for systems that can be trained to learn is machine learning
One way to use machine learning is by simulating learning in the brain
This is what is called neural networks
It is important to understand that learning systems are not programmed in the task they perform, they are feed data and trained
Machine Learning
A computer program is said to learn from experience E with respect to some class of tasks T and performance measure P, if its performance at tasks in T, as measured by P, improves with experience E
Machine Learning
Breakthroughs in computer performance (GPUs), algorithms, cloud computing and big data, has finally created an environment where neural networks - systems that learn have become a reality
The ideas of learning systems came very early but failed to become practical
Fraud detection Web search results Real-time ads on web pages and mobile devices Text-based sentiment analysis Credit scoring and next-best offers Prediction of equipment failures New pricing models Network intrusion detection Pattern and image recognition Email spam filtering
Application
Types of AI Today
Cognitive Systems Neural Networks and Deep learning Generic Algorithms Artificial General Intelligence (AGI)
Cognitive SystemsCognitive systems are knowledge based systems
They are fed with information and can observe and learn
They work on sort of act – learn – loop
These systems are human architected as they need to be feed with lots of information.
IBM Watson is an example of cognitive systems.
Neural Networks and Deep learning
Old technique developed in the 1950s and 60s
Today, with new unprecedented scale both of data and computing power these systems they have new properties that allow them to solve problems previously very difficult, like image labelling
These neural networks work like this: the network is made up of neurons and connectors. They have input layer, hidden layers of neurons and connectors, and output layer.
Generic Algorithms
Work similar to biology’s natural selection or survival of the fittest
To begin with the algorithms are given a task and to solve it they will try random solutions
The outcomes of these are then evaluated by a fitness function and some of the outcomes will perform better than others. The better ones are upgraded and the worse are downgraded and this is then repeated.
Artificial General Intelligence (AGI)
What AGI is about is unsupervised learning which is one of the hardest problems in AI
The other types of AI use labelled data and a fitness function
It can be instructed on how to improve
In general, AGI has shown little progress, except for some isolated cases like game cases
State of AI Today
We are in an AI Boom
Google, IBM and other tech giants started to develop more solutions, such as pattern recognition, interpretation of medical images, visualizing, recognizing objects in images, controlling cars and robots to name few.
Google has TensorFlow, an Open Source Software Library for Machine Intelligence
Machine Learning Platform
Now platforms are becoming available
Amazon has Amazon Machine Learning
Microsoft is providing machine learning as part of Cortana Analytics Suite - Microsoft Azure Machine Learning
Facebook has FBLearner Flow
Human intelligence
Artificial intelligence We are here
Inte
lligen
ce
Time
Is this the path to Machine Intelligence?