Stanford Freshman Seminar: Traveling the Information Highways: ENTertainment and EDUcation

Maps, Encounters, and Directions

Master copy on Varese
Spotty draft, from earlier material 15Jan98, updates 7Mar1998, Jan 2003, 25 jan 2004, plus much unedited copied material.
This material is

©Gio Wiederhold and CS99I students, Stanford University, 1998, 2003.

Previous chapter: Browsing - Next chapter: Digital Libraries

Chapter: Entertainment and Education

We are all teachers, and we always teach what we know [Shirley MacLaine (actress), on receiving the Cecil DeMille Lifetime award, January 1998]

ENTEDU.Intro

We combine entertainment and education in one chapter because of the great similarities on objectives, methods, and technology in both domains. Differences do exist, but we see that the electronic highways will accentuate the similarities.

Paper by Andrew Dickson

ENTEDU.Intro.knowledge

When we deal with education and entertainment three terms are used frequently, and often not precisely. Since they are crucial to further discussions, we will distinguish them throughout this book, and define them now.

· Data: recordings of factual observations. They should be verifiable by measurements, unless the state of the world has changed. Historical data should be time-stamped to avoid confusion.

Examples:

• The Gates Computer Science building at Stanford has a red roof..
• The CS99I freshman seminar in Winter 1997/1998 has 8 students.

· Knowledge: general rules about the world. They are learned, perhaps by looking at many data elements, or given as explicit constraints..

Examples:

• Stanford buildings have red roofs.
• Freshman Seminars have are limited to 16 students.

· Information: combinations of relevant data and knowledge, useful to customer.

Examples:

• for a pilot: There is a clump of red roofs, If I fly East I will find Palo Alto airport.
• for freshman: I can still enroll in CS99I.

It is not surprising that the boundaries are often blurred. This book, for instance, is strictly data by this strict definition until it is read and acted on by a reader. A reader who now creates web pages according to the prescription in the HTML note uses it as information, as was the intent of the author, but for the author it was just data. The transmission of data from a source to a receiver who is not aware of those facts is an important method in creating information. The reader also had to contribute knowledge to convert this data into information: how to locate and access the particular web page, and background to know what terms as heading, list, etc. mean. A model is hence that we convert data to information, as sketched here.
http://www-db.stanford.edu/pub/gio/CS99I/figures/datatoinf.gif
Fig. Data-to-Information: Converting Data to Information

.If the information is used for an action, as enrolling in CS99I, the state of the world is changed, and hence the observed data have to be updated. We now have a data loop, As we handle more data we create useful abstractions, perhaps that CS freshman seminars are rarely filled. This rule then becomes knowledge for later reuse, or for transmittal to fellow students. Knowledge, being general, is more compact than data: the list of all CS freshman seminars.

Teaching is concerned with transmission of knowledge. In order to substantiate the knowledge, teachers often use factual examples, since those can be verified by the students. Knowledge is powerful, but often less precise. There are some buildings at Stanford which do not have red roofs, but that will not confuse the pilot as long as it is largely true, and a distinction with respect to other neighborhoods in the San Francisco Peninsula.

http://www-db.stanford.edu/pub/gio/CS99I/figures/dataloop.gif
Fig. Data-Loop: Updating of Data and Knowledge

.

ENTEDU.Support

One difference in Entertainment and Education been in support, and that has made a difference in presentation:

· Education is largely supported by taxes (see article from Economist, copied below)

· Entertainment is largely supported by admission fees.

Both domains seek support of benefactors, since taxes and receipts often fall short of costs. Entertainment, when it can claim benefits for the public good, as Public Broadcasting (PBS) and symphony orchestras, seeks tax support, and many of the best educational institutions levy substantial fees.

In the electronic world collecting fees becomes more difficult, we will deal with these business aspects in the Chapter on Electronic Commerce.

ENTEDU.History

Both entertainment and education have a long history, and we will make only some points that relate to the electronic highways.

Expectation for education are changing. Students are used to the pace and soundbytes of TV. Multimedia presentation are expected. Paper text books are boring, and getting to big.

The information technology that will be used for modern education will adopt many concepts from entertainment: reliance on graphics, interaction, instant replay, multiple paths to reach goal, etc. [Brutzman:97]. The initial uses of the Internet in education are simple, and similar to access requests by scientists and consumers [Perrochon:96], but material specific for on-line education is being developed by a variety of places [diPaolo:99]. The tool providers will have to reengineer the tools developed for creating entertaining games to make them suitable for authoring by teachers. Much work is required by educators to present and maintain educational material [VernonLP:94]. The market for an educational product has to be larger than a single classroom to be viable. Such markets now exist in industrial training, where students cannot be brought together at one time in one place. Acceptance of distributed high-quality academic material will start where colleges cannot cover all topics of interest to their students, and will broaden as successes are attained.

Study material is likely to adopt hypertext formats. Section would be presented only if the student indicates that more information or background on a topic is desired. There may be sections that would be blocked untill the student has read a pre-requisite section or answered questions indicating competence.

ENTEDU.History.minstrels

Distribution of information by wandering individuals.

ENTEDU.History.colleges

Distribution and generation of information by groups.

ENTEDU.History.broadcasting

Distribution of information by dissemination.

Teaching

Shows

ENTEDU.Business

Education is an enterprise that affects a large fraction of our population (20%?) and that fraction is increasing as society and technology change and require frequent updating of one's skills. Even the assumptions about a student's future options, made when entering college, are likely to have been overtaken by the time of graduation.

Education

The classical model of education and training is based on direct communication between a teacher and the student or students. To be effective this has demanded small groups of students, physically collocated with the teacher. Recent adoption of TV broadcasts has effectively enlarged the size of the class that can be addressed by one teacher by electronically expanding the classroom to include the locations served by the receiving sites. Despite the significant advantage TV has brought, the model is fundamentally unchanged: one teacher providing instruction to a number of students in real time. The use of video recording technology allows a student to "attend class" in a delayed time mode, but the teacher-classroom model remains essentially intact. There have also been other successful applications of technology to the education and training process, but the overwhelming majority of ETLL providers remain entrenched in the classic model, or introduce slight variations to it.

Stanford, through its SITN program, has provided remote education, transmitted from Stanford to class rooms in industrial sites. This program has been has been successful and effective when the sites receiving had a tutor assigned. Problems are due to delays in getting homework in or out -- but homework is still based on paper. There is an opportunity for remote students to ask questions in real-time

How will this model work for other colleges?

The potential loss of individual teaching opportunities is already causing reactions in some teacher organizations who rightfully fear that acquisition of costly material from remote institutions will diminish their interaction with students. Such concerns will delay the adoption of information technology in education but cannot halt it. Education today is based on information in books, although five hundred years ago some reputable authorities did not expect printing to be useful [Hibbitts:96]. Electronically mediated information is likely to become the principal carrier of information for education, and effective teachers will learn how to manage and exploit it. How these capabilities will change the process and structure of education is hard to predict, but it seems unlikely that another hundred years hence much time will be spent by teachers standing in front of a class and holding forth.

Examples

University of Phoenix.

More students than any other school.

Not limited by physical constraints.

Profitable (president bought $30M house in San Francisco)

Training

[Based on a conversation with a manager at H-P, 2000.]

I wanted to respond to your recent message regarding on-demand education. Based on our discussions with engineering and education/training managers at SITN's member companies you're right on target. What we're hearing is that practicing engineers, and the companies that pay for their education, want "control" of the teaching and learning. They want control over the place and time (ideally at the desktop or even at home), pace, and even the scope and sequence of the material --- and not be constrained by the barriers imposed by the traditional on-campus class. If you wish, I can send you an outline of our assessment of the industry environment and engineering education.

The idea of smaller increments of instruction is something SITN has been working on with Stanford faculty in the development of non-credit short courses. These are programs that average five hours in length and are typically broken into one hour increments. The programs are taped and offered as a five hour course on satellite and on video tape. In the future these courses could be divided up and offered as modules available on-line from video servers. The products (including regular 30 hour courses designed to be broken into stand alone modules) could also be converted into CD-ROMs with the entire course indexed for easy access. In fact, we have developed a CD-ROM demo of a repurposed engineering class that might serve as a model for future development.

Stanford faculty have offered about 15 short courses over the last 20 months in a range of engineering disciplines. Examples include: C++ (Cheriton), Digital Circuits (DeMicheli), T-CAD (Dutton), Composites (Tsai), Cryptography (Hellman), Turbulence Modeling (Bradshaw) and Design for Assembly (Barkan). Our target is to have at least one of these a month, eventually ratcheting it up to three a month using both Stanford faculty and distinguished industry experts.

As Jeff Ullman alluded to in his recent note, a few of us (Jeff, Fouad Tobagi, Dale Harris, Dwain Fullerton et al.) have been trying to figure a way to run an experiment to get some of the School's televised classes and short courses available to customers on-demand using video servers. With the real possibility of some resources to get this idea started (and Gibbons endorsement), I suggest those who are interested gather to talk about next steps. It would probably make sense to have a few potential industry customers in attendance at one of these sessions so that they can provide a reality check about what we have in mind -- they are the ones who are going to pay to receive the programming. For example, Hewlett-Packard's corporate engineering education group is currently delivering our 250 courses to the workstation as a live signal, but they are very interested in having a menu driven "pull" system. Their idea is to have SITN repurpose existing video product into smaller increments so that an H-P engineer can "pull" modules of instruction or information when and where needed. I know H-P would have an interest in discussing this concept with Stanford engineering faculty. Since H-P represents over 50% of the School's external engineering education business we should listen to their suggestions. H-P and other SITN companies might also have resources they wish to add to the mix.

A significant computational challenge is presented by the need to more carefully define the skills needed to perform real job functions, assess the level of relevant skills in employees and applicants, and define the course modules necessary to make up the difference. The skill assessment work that is being done today is still largely a craft, performed by highly trained professionals without significant technological support. As a result it does not scale to the nationwide application that is necessary in order for the new, skill based ETLL model to become a reality. These processes need to be automated, which will require programs that are capable of dealing with large amounts of information and perform processing that can deal with and resolve the ambiguities that still plague the "soft sciences." Probabilistic rather than deterministic processing will be necessary, suggesting the need for Artificial Intelligence or "Fuzzy Logic" applications. To automate and scale the current "expert" processes will demand both sophisticated software and high capacity, high speed computational hardware.

ENTEDU.Functions

[[copied material, to be massively edited for content ]]

NATIONAL CHALLENGE: Education, Training and Lifelong learning (ETLL) Education "White Paper"

The Challenge.

The enormous and accelerating advances in technology in the past century have brought many benefits to nearly everyone, particularly in the industrialized nations. While this advance can be expected to continue unabated, in order to accrue a corresponding increase in benefit, it is necessary to concentrate more specifically on the one unchanging key component of every system: The human being. The fundamental human characteristics of memory capacity, input and output bandwidths etc., are essentially unchangeable. What can and must be improved and maintained is the skill set with which each person is equipped. This requires more effective education at all levels, improved job specific skill-based training and comprehensive accessibility of Lifelong Learning resources that will enable each person to maintain and develop the skills necessary to live and work successfully in an environment of constantly changing demands and opportunities.

The classical model of education and training is based on direct communication between a teacher and the student or students. To be effective this has demanded small groups of students, physically collocated with the teacher. Recent adoption of TV broadcasts has effectively enlarged the size of the class that can be addressed by one teacher by electronically expanding the classroom to include the locations served by the receiving sites. Despite the significant advantage TV has brought, the model is fundamentally unchanged: one teacher providing instruction to a number of students in real time. The use of video recording technology allows a student to "attend class" in a delayed time mode, but the teacher-classroom model remains essentially intact. There have also been other successful applications of technology to the education and training process, but the overwhelming majority of ETLL providers remain entrenched in the classic model, or introduce slight variations to it.

The demands of true lifelong learning for all citizens can only be met by application of digital processing and advanced communications technologies that will enable substantive, individualized training and education to be delivered on-demand, at affordable costs, anywhere in the country and any time of the day or night.

The Task

It is time to replace the classical ETLL model with one that allows the individual to pursue new knowledge and skills in a manner that places the location, content and timing of the process fully under his or her control. The power and rapidly increasing availability of computers and digital communications now make it possible to realize this new model. Because the classical model is so entrenched, the change to a new concept will encounter scepticism and resistance on the part of many potential users as well as some established education and training institutions. Overcoming these obstacles will require careful planning, combined with compelling pilot implementations that clearly demonstrate effectiveness, usability and economic advantage of the new approach. Effective planning to bring HPCC technology to bear on the ETLL challenge will require partnership with practitioners of the "soft sciences" who are expert in the areas of job and personal skill assessments (industrial psychologists). It will also be necessary to bring technological expertise together with educational psychologists and practicing teachers and trainers, both to learn how best to apply technology to the teaching/learning process and to assist the practitioners in developing confidence in its efficacy. There is also a major need for the technological community to reduce the amount of specialized knowledge necessary to access and employ applications designed to benefit education and training professionals and individuals pursuing independent learning objectives.

Role of computation, information processing, storage and communications in modernizing Education, Training and Lifelong learning.

Computation: A significant computational challenge is presented by the need to more carefully define the skills needed to perform real job functions, assess the level of relevant skills in employees and applicants, and define the course modules necessary to make up the difference. The skill assessment work that is being done today is still largely a craft, performed by highly trained professionals without significant technological support. As a result it does not scale to the nationwide application that is necessary in order for the new, skill based ETLL model to become a reality. These processes need to be automated, which will require programs that are capable of dealing with large amounts of information and perform processing that can deal with and resolve the ambiguities that still plague the "soft sciences." Probabilistic rather than deterministic processing will be necessary, suggesting the need for Artificial Intelligence or "Fuzzy Logic" applications. To automate and scale the current "expert" processes will demand both sophisticated software and high capacity, high speed computational hardware.

One of the principal advantages that digital technology can bring to the ETLL process is the ability for the student to "learn by doing" in simulated environments. Simulations have the advantage that they are safer and less expensive than the actual experience, and can be more controlled in ways that will optimize the learning process. Creating simulations of this quality will place large computational demands, particularly in those cases where the simulation itself must be capable of being reproduced on platforms that are inexpensive enough to be commonplace in the home or office.

Information processing: The challenge confronting information processing stems largely from the scale and variety of the ETLL problem. Skill requirements for virtually any job need to be kept current and readily available to industry and individuals. Each individual needs to be able to assess his or her own skills in terms of a standard metric and use that assessment to make personal career or ETLL decisions. These personal data must be afforded security such that they are accessible only to those with a legitimate right to them, yet must be easily shared among those with a valid need as determined by the data owner. The interface between the individual and stored courseware modules (e.g., simulations) must facilitate the identification and presentation of the desired ETLL modules with minimal demand for specialized technical knowledge or skills. The Information processing system must also keep track of the ETLL records of millions of individuals and maintain currency of skill databases and skill-based ETLL course modules stored in simultaneous multiple locations on the network. The information processing system and must enable course modules to be selected from competitive offerings, acquired and paid for electronically (using electronic commerce processes).

Communications: A new, technology based model of ETLL will require multiple concurrent access to training opportunities from anywhere in the country. This can only be accomplished through ubiquitous and reliable broadband communications, that can move the educational and training materials to the student on demand. Because some of these materials may consist of simulations or video segments, the bandwidth requirements, at least in the users local environment. will be large (several megahertz), however, movement of the modules into storage in the local environment could occur at slower rates than those required for presentation to the student. In many learning situations and for many learners, peer interaction is a valuable part of the learning process. Consequently the communications systems must support one-to-one and one-to-many communications, both in real time and delayed mode, at the option of the student (or mentor/teacher, where appropriate). In most cases, simple audio or email style communications will probably suffice; however, there will probably be occasions when true video conferencing will be essential to the learning process, and must consequently be supported. For certain users and in certain circumstances it can also be anticipated that mobile or portable communications will be used in ETLL applications. Storage: Large amounts of information storage capacity will be required to house the numerous multimedia course modules that will be necessary to realize the vision of ETLL available to anyone anywhere at any time. For reliability and to avoid frequent overload of long distance telecommunications, several large "warehouse" storage facilities serving population centers or geographical regions will probably be needed. The course modules in the various servers will largely be the same, and it will be necessary to ensure that any updates to course modules are made to all copies of those modules regardless of the location of the servers housing them. (It can be anticipated that regional warehouses will contain some course modules that are only of interest to the local population and which would consequently not be replicated in other warehouses.)

Relationship of this HPCC/IITA National Challenge activity to those of other agency activities.

High level interest in both the legislative and executive branch has stimulated vigorous activity in the ETLL area in virtually all government agencies: The Office of Science and Technology Policy Special Assistant for Education and Training is actively promoting technology based training and coordinating the activities of interdisciplinary working groups and committees The Department of Education is logically the focus of much of the current effort, particularly in the K-12 area. The establishment of a specific office for Educational Technology demonstrates the strength of their commitment, as does the intensity of their efforts to assist the State education agencies in adopting modern technology in their school systems.

The Department of Labor, in line with the creation of the Skill Standards Board, has begun a process that will allow trainers, educators and individuals to specifically identify the skills required for success on the job.

The Department of Commerce, through the NTIA, is currently in the process of reviewing grant applications in anticipation of awarding millions of dollars worth of grants to support technology planning activities in the various states.

The National Science Foundation continues to support educational research and development through grants and has also supported development of technology based courseware in the Math and Science areas.

NASA actively supports K-12 education, placing resource material on the Internet for access by students doing school assignment and projects.

Like the NASA, the Department of Energy maintains instructional material available to students over the Internet. As with the NASA offerings, The DOE materials receive high praise from the teachers whose students have been able to use them.

The Department of Agriculture fosters distance learning through the Land Grant Colleges, and Americans Communicating Electronically. The Department of Defense has been very active in use of simulation and networks for training purposes, and is in dialogue with non-defense agencies with the goal of making appropriate learning modules available to the civilian population.

NSA, in concert with other federal agencies, has joined with a group of industrial onsumers of job related training, forming a consortium to pilot applications of Job Skills Analysis leading to certified technology based training modules to develop or enhance needed skills.

Included in this effort is the establishment of the non-profit American Training Standards Institute to coordinate and manage the skills assessment and course module certification processes.

ENTEDU.Functions.broadcast

ENTEDU.Functions.multicast

ENTEDU.Functions.on-demand

Here's a response I sent to Paul Losleben last week around the "education on-demand" concept and some approaches we are working on to deliver Stanford programming to industry. I've had a response >From Ullman, Tobagi and Harris and will be following up with them in the near future.
Andy DiPaolo
Assistant Dean, School of Engineering
--------------------------------------------------------------------
Paul,
I wanted to respond to your recent message regarding on-demand education. Based on our discussions with engineering and education/training managers at SITN's member companies you're right on target. What we're hearing is that practicing engineers, and the companies that pay for their education, want "control" of the teaching and learning. They want control over the place and time (ideally at the desktop or even at home), pace, and even the scope and sequence of the material --- and not be constrained by the barriers imposed by the traditional on-campus class. If you wish, I can send you an outline of our assessment of the industry environment and engineering education. The idea of smaller increments of instruction is something SITN has been working on with Stanford faculty in the development of non-credit short courses. These are programs that average five hours in length and are typically broken into one hour increments. The programs are taped and offered as a five hour course on satellite and on video tape. In the future these courses could be divided up and offered as modules available on-line from video servers. The products (including regular 30 hour courses designed to be broken into stand alone modules) could also be converted into CD-ROMs with the entire course indexed for easy access. In fact, we have developed a CD-ROM demo of a repurposed engineering class that might serve as a model for future development. Stanford faculty have offered about 15 short courses over the last 20 months in a range of engineering disciplines. Examples include: C++ (Cheriton), Digital Circuits (DeMicheli), T-CAD (Dutton), Composites (Tsai), Cryptography (Hellman), Turbulence Modeling (Bradshaw) and Design for Assembly (Barkan). Our target is to have at least one of these a month, eventually ratcheting it up to three a month using both Stanford faculty and distinguished industry experts. As Jeff Ullman alluded to in his recent note, a few of us (Jeff, Fouad Tobagi, Dale Harris, Dwain Fullerton et al) have been trying to figure a way to run an experiment to get some of the School's televised classes and short courses available to customers on-demand using video servers. With the real possibility of some resources to get this idea started (and Gibbons endorsement), I suggest those who are interested gather to talk about next steps. It would probably make sense to have a few potential industry customers in attendance at one of these sessions so that they can provide a reality check about what we have in mind-they are the ones who are going to pay to receive the programming. For example, Hewlett-Packard's corporate engineering education group is currently delivering our 250 courses to the workstation as a live signal, but they are very interested in having a menu driven "pull" system. Their idea is to have SITN repurpose existing video product into smaller increments so that an H-P engineer can "pull" modules of instruction or information when and where needed. I know H-P would have an interest in discussing this concept with Stanford engineering faculty. Since H-P represents over 50% of the School's external engineering education business we should listen to their suggestions. H-P and other SITN companies might also have resources they wish to add to the mix..

ENTEDU.Functions.feedback

ENTEDU.Functions.collaboratory

Gio, we describe below some ideas regarding data base tools to capture the value added to data during a collaboration session involving a distributed team supported by electronic collaboration tools-the collaboratory. These ideas would augment our ongoing collaboratory development, however, data base activities are not now supported. I will look forward to your reaction and guidance as to how to proceed. Best wishes, and warm regards, Bob

ENTEDU.Technology

Both entertainment and education require high transmission bandwidth. There are two components to this issue:

1. The material is voluminous, and often includes images and video clips

2. The number of recipients is large

ENTEDU.Technology.multimedia

The material we deal with in entertainment and education contains text, images, and today often voice and video as well. Smell is still rare. For each of those data representations there are a variety of technological issues and solutions. We will deal with them individually here, but must keep in mind that in the end an integrated presentation is desired, where text, images, voice, and video are synchronized. Achieving such a synchronization in the shared communication lines of the Internet is still a major challenge.

Multimedia in Japan Today and Tomorrow
By Iwasaki Ieo

The mass media in Japan is placing heavy coverage on information about multimedia. Seminars, symposia, and expositions on this theme are being held with incredible frequency and numerous books, ranging from educational primers to fairly specialized works, are being published. To say that Japan is now in the midst of multimedia fever is indeed no exaggeration. [Hyperbole and exaggerated extrapolations are everywhere, as in the US. The multimedia excitement significantly overlaps that associated with the Internet, on which four books appeared in bookstores here last week. Several ministries are competing for major control of this new technology, including the Ministry of Posts and Telecommunications (MOPT, but usually abbreviated MPT), and the Ministry of International Trade and Industry (MITI), but also including others such as the Ministry of Construction, and the Ministry of Education (Mombusho). Japan is not alone in Asia in getting involved in multimedia. Recently, the Korean Daewoo Electronics Company announced that it would invest US$2B over the next ten years in various multimedia services and equipment and will get involved in cable and satellite broadcasting, the production of compact discs and CD-ROM software, electronic printing, films and theaters, with an intention of being one of the world's leading multimedia companies by 2015. Daewoo's plans begin with the establishment of the Daewoo Cinema Network for CA-TV, and will follow with satellite broadcasting, HDTV, etc. DKK] The video game equipment sector is also looking ahead by putting a steady stream of CD-ROM players, designed for the future multimedia age, on the market. Companies are employing animation compression technologies in the production of all of these game machines, whose high functionality, designed for the multimedia era, is used as a selling point.

Appliance manufacturers are also selling home video CD players with simple, built-in interactive functions while personal computers equipped with CD-ROM drives, or, "multimedia PCs," that have recently entered the market are becoming the mainstream type of equipment.

The direct impetus for this multimedia boom was US Vice President Al Gore's announcement in September 1993 of an action plan for the construction of an "information superhighway" that would use new infrastructure to raise educational, medical, and other social welfare levels by the year 2010. This had a tremendous impact on Japan. The average person had considered multimedia something indistinct and in the distant future, but it had suddenly begun to take shape in Japan.

DAVID L. TENNENHOUSE
Telemedia, Networks & Systems Group
Laboratory For Computer Science
Massachusetts Institute Of Technology
The first wave of media applications, i.e., those that simply copy and store multi-media objects, will be followed by a second wave of computation-intensive applications-that actively process the media-based information. These applications extend the requirement for `video to the desktop' to a more general requirement for `media to the application'.

The ViewStation architecture embodies a software-oriented approach that supports this `media to the application' paradigm. Our programming environment makes the raw media data, e.g., the actual video pixels and audio samples, accessible to the applications.

We have derived a set of architectural guidelines and have constructed an integrated system that supports media-intensive applications. The principal components of our `stack' are:

· * the VuNet, an ATM-based Desk Area Network;

· * the VuSystem, a media programming environment; and

· * COMMAA, a suite of media-intensive office applications.

[[end of copied material]]

ENTEDU.Technology.compression

Need

[[Copied material, to be massively edited]]
Garson emphasized the difficulties of transmitting images over the networks-a full page image, uncompressed, could take 2 hours to transmit at 1200 baud, 14 minutes at 9600, 2.4 minutes at 56000, and 5 seconds on a T1 line.

MPEG

BERND WOLFINGER
Hamburg University, Germany
Computer Science Department
berndw@icsi.berkeley.edu
"Efficiency of PET and MPEG Encoding for Video Streams: Analytical QoS Evaluations"
A promising solution in the transmission of video streams via communication networks is to use forward error control in order to mask some of the transmission errors and data losses at the receiving side. The redundancy required, however, to achieve error correction without retransmissions will consume some transmission capacity of a network therefore possibly enforcing stronger compression of the video stream to be transmitted.

In this talk we introduce analytical models which allow us to determine the expected frame loss probability of MPEG encoded video streams assuming communication via constant bit rate (CBR) virtual circuits with data losses and/or unrecoverable transmission errors. The models can be used to compare the quality-of-service (QoS) as observed on Application Layer for encoding schemes without and with forward error control, possibly making use of different prioritization of transmitted data units (in particular applying PET encoding algorithm as designed at ICSI). The talk covers preliminary results and is conceived as a forum for critical discussions on the approach chosen

.

ENTEDU.Technology.redistribution

MIT making all its courses on-line available

<<add reference>>

Geometric constraint solving

We are pleased to announce the availability of an electronic primer on geometric constraint solving developed for the ONR research community. The primer is an electronic book available on world-wide-web through XMosaic. It can be read following four predefined "tours", or following hyperlinks. The primer also contains instructions on downloading the constraint solver to be run locally, on Sun workstations, as well as the capability to run the constraint solver locally at Purdue. The URL access is
http://www.cs.purdue.edu/homes/pjv/book/intro.html

The tours are as follows:

1. 1. Overview The overview tour provides an introduction to geometric constraint solving with emphasis on our approach to constraint solving.
2. 2. Tutorial The tutorial tour is a hands-on guide to using our two-dimensional constraint solver.
3. 3. Theoretical Foundations The theoretical basis for our constraint solver is detailed in this tour.
4. 4. Implementation This tour presents the theoretical and technical aspects of our constraint solver which are necessary in order to be able to implement a constraint solver similar to ours.

There is also a bibliography arranged alphabetically by author.

[Christoph M. Hoffmann, Purdue University, <cmh@cs.purdue.edu>
Pamela J. Vermeer, Washington and Lee University, <pvermeer@wlu.edu>]

Radiology

Continuing Medical Education (CME) credits are now available online!
The University of Washington School of Medicine has designated the digital teaching file on the UW Radiology Webserver for credit hours in Category I of the Physicians Recognition Award of the American Medical Association.
The URL for this server is: http://www.rad.washington.edu/
Other learning modules are currently available on this server and will also be available for CME credit soon. This server includes the following items:

1. 1. Radiology Teaching File
2. 2. Anatomy Teaching Modules
3. 3. Radiology Exhibits from UW
4. 4. Information on UW radiology residency and fellowship programs
5. 5. Image processing software written by UW faculty

For further information, contact:
[Michael Richardson, M.D. <mrich@u.washington.edu>.]

Harvard Business School

To create course material they are doing Video logging of Harvard Business school material. The result is then made available for remote intercative learning.

Video logging breaks a video tape into logical segments (by speaker, change of scene, change of topic, intermediate music, etc.), and then creates index entries for each segement, so the the information in that segment is rapidly accessible. For each segment a key frame is identified and shown. Information extracted includes words from speech-recognition, , close-captioning, face recognition, speaker voice identification, etc.

faculty involved there [Larry Bouthillier (*contact?); Sandy Pentlam (sp?)]. Examples found by search on `stock', 'Cyberposium 99'.

New School for Social Research - University Cybercampus, New York City

ENTEDU.Technology.sharing

Advanced Multimedia Database Tools to Support Distributed Scientific Team Analysis and Collaboration

INVESTIGATORS
Elke A. Rundensteiner
Assistant Professor
Software Systems Research Laboratory
Electrical Engineering and Computer Science Dept.
The University of Michigan
Ann Arbor, Michigan 48109-2122
Phone: (313) 936-2971
Fax: (313)
Email: rundenst@eecs.umich.edu

Robert Clauer
Research Scientist
Atmospheric, Oceanic and Space Sciences Department
Ann Arbor, Michigan 48109-2143
Email: clauer@pitts.sprl.umich.edu

Jason Daida
Terry Weymouth
Atul Prakash

ABSTRACT:
ADVANCED MULTIMEDIA DATABASE TOOLS TO SUPPORT DISTRIBUTED SCIENTIFIC TEAM ANALYSIS AND COLLABORATION

We would like to design, implement, test, and distribute a software tool-kit that supports data analysis by a geographically distributed science group.

Our vision, and the purpose of our proposal, is to enable a distributed team of scientists to work together with their data in a more productive fashion. The distributed scientific team will be supported by an emerging electronic infrastructure called the National Information Infrastructure (NII) and new object-oriented multimedia database technologies. Building upon collaboration tools being developed under separate NSF support for ground-based science (NSF Upper Atmosphere Research Collaboratory, or UARC), we propose to leverage off this NSF project to implement a distributed team collaboration facilitator.

To create the facilitator, we will design and implement the software technology for researchers to jointly interact with data, add annotation and discussion that can be accumulated, retrieved, edited, added to, using distributed multimedia database technology. Key technologies will include distributed database tools to support archiving collaboration sessions, as well as the retrieving and updating of these sessions. The proposed suite of software tools would thus support the team analysis process from initial data collection all the way through the publication of new results and knowledge.

We will utilize the existing Upper Atmospheric Research Collaboratory (UARC) collaboration tools developed through NSF support. We will augment the UARC collaboration tools with object-oriented, multi-media, data base tools to capture information from collaboration sessions. We will design and implement the software technology for researchers to jointly interact with data, add annotation and discussion that can be accumulated, retrieved, edited, added to, using distributed multimedia database technology. Key technology will include distributed database tools to support archival of collaborations sessions and retrieval and update of these sessions. The proposed suite of software tools would thus support the team analysis process from initial quick-look data all the way through publication of new results and knowledge.

By providing a basic collaboration framework, we envision each member of a research team to be able to:

1. (1) link electronically through their workstations

2. (2) utilize shared data display windows

3. (3) rapidly prototype a new view of the shared data

4. (4) exchange email and voice dialogue

5. (5) share drawing tools

6. (6) make annotations upon the data.
By providing multimedia database tools within this framework, we now envision members of this research team to also be able to:

7. (7) store and save an entire collaboration session

8. (8) fast-forward, rewind, skip-forward, skip-backward, pause and play any previously stored collaboration session.

9. (9) replay any previous collaboration session within the context of another collaboration session.

10. (10) search, browse, and retrieve content from within any stored session.

To accomplish this level of functionality for the database tools, we propose to develop a new type of dynamic metadata that should be saved in a multimedia object-oriented database system. Note that our usage of the term metadata does not simply refer to descriptive data about the raw scientific data (e.g., netCDF, HDF), but also descriptive data about the process-context in which the raw scientific data appears (i.e., context-sensitive metadata that observes the temporal and distributed relationships between multimedia artifacts). Such multimedia objects must be synchronized with the scientific data being investigated, in addition to establishing possibly complex interrelationships among different types and groups of multimedia annotations.

Members of the research team will be linked electronically through their workstations, utilizing shared data display windows, typed and voice dialogue, shared drawing tools and annotation upon the data. The dialogue, discussion, annotation and drawings which result from such collaboration sessions form a new type of dynamic metadata which should be saved in a multimedia object-oriented data base system. Note that our usage of the term metadata does not simply refer to descriptive data about the raw scientific data, as commonly assumed in the community, but rather we are targeting truly diverse multimedia data including hand drawn sketches representing graphical interpretations of phenomena observed on the scientific data, general conversations about the type of scientific observations or even the scientific process in general. Such multimedia objects must be synchronized with the scientific data being investigated, in addition to establishing possibly complex interrelationships among different types and groups of multimedia annotations.

In short, we will employ state-of-the art digital library technology to achieve this level of collaboration support for scientific processes, consisting of the following components:

· A. Collaboration support, including on-line shared communication, dialog, annotations, session archives and replay.

· B. Shared database views for update and coordination of scientific data, collaboration-related information, and display and visualization type of session information.

· C. Distributed object-oriented multimedia collaboration database, covering data model development, data synchronization, and query languages.

· D. Synchronization of scientific data with its multimedia interpretations, indexing, browsing, and retrieval of scientific data based on value-added expert interpretations;

· E. Distributed workstation environments for test bed development.

IMPACT

The proposed research into constructing distributed multimedia object-oriented database tools supporting scientific collaborations will clearly make a major impact on the ease with which scientists- distributed geographically among several institutions-can advance in their scientific interactions to generate publications of the studied data. Facilitating teams of investigators to collaboratively study science data should increase their productivity. More importantly, however, while the collection of 'raw' scientific data is important in general, the collection of interpretations of such scientific data by 'the' experts will be a true valued-added asset. Note that the augmentation of the archived science data sets with valued-added interpretations generated by 'the' experts of the data will be a natural by-product of their scientific studies, rather than requiring a pain-staking effort by the scientists in documenting their findings. Indeed, documentation tasks often receive a low priority because they are tedious, even though such tasks are often deemed important to the overall scientific inquiry.

Furthermore, given that such annotations and interpretations are typically of several orders of magnitude smaller in quantity in comparison to the amount of actual data, it would be much more feasible to successfully interrogate and retrieve information based on these interpretations. In fact, these interpretations will typically focus on 'interesting' data sets, thus providing key pointers to meaningful features that would otherwise be buried in a sea of information. The proposed multimedia database will be a key technology in extracting truly useful information from scientific investigations. It will provide support for 'replaying' of previous scientific sessions, which would allow for annotating or revising previous interpretations with new information. Furthermore, this will bring scientific data and the scientific process itself into a format so that it can be utilized for educational purposes to demonstrate the scientific process involved in studying and learning from data.

As noted above, this work will be undertaken in a testbed environment, utilizing the xisting UARC collaboratory testbed. Ultimately, however, the results of this proposed effort will have impact far beyond just the space science community. Many science team in all disciplines could benefit from the technology that we propose to develop. The generic quality of this technology could impact distributed teams who must work together over data in most all scientific disciplines, in engineering, in business, and in education. (For example, students could learn about the process of satellite image interpretation by collaborating with students at other distant locations to obtain "live" ground truth information.) While we are proposing a testbed in a scientific context, we feel that the impact of the technology will be much broader, affecting all collaborating teams in all manner of activity.

[[end of copied material]]

ENTEDU.Technology.compression

Compression reduces the volume of data to be transmitted of the networks and stored on the storage devices by taking advantage of the inherent redundancy in the representation of information that we use. Redundancy in writing allows us to understand a sentence in the presence of some typos or smudged print. Redundancy in speaking allows us to understand a message even when a slamming door makes causes us not to hear a word. A car ad with a staple in the center still conveys its message. We can follow a film even if distracted for a minute. However, for each of the scenarios we can make up instances where the loss would be significant. For each of these media we have to consider what is truly redundant and what is of marginal benefit to the intended receiver.

Even with increasing bandwidths, there is still a need for compression. Compression techniques have been developed to provide compression ratios varying from as low as 10:1 to as high as 60,000:1. Compression can be applied to for texts, sound, images (ranging from line drawings to animation or moving pictures) and video.

Ideally we do not want to loose any information in the sequence Compression, Transmission, Storage, Transmission, Decompression. Lossless compression is essential when even one bit change can make a crucial difference in the result say an equation E= mc^2 is changed to E= mc^3. Similar precision is needed for musical notation. In general text has to be compressed without loss, because its redundancy is low (~50%). Formatting information associated with text may have more redundancy. Voice, images, and video contain much redundancy, so that the loss of a few bits may not be noticed, even though the receiver may still feel uncomfortable about any loss. A radiologist, receiving an X-ray, will be legitimately concerned about any loss. However, images used for entertainment and education are deemed to be less critical, so that here lossy compression dominates, since much greater compression ratios can be achieved.

Compression does require computation capability. Since we expect that compression will be less frequent than decompression; material is read more often than written, most schemes put much computational effort into compression, and set the results so that decompression is fast, preferably as fast as the data can be received. The most powerful compression methods investigate an entire document for redundancy, create tables of recurring patterns, and then transmit first those tables, and then the skeleton of the document, where each occurrence of a pattern is replaced by a reference. The delay implicit in this process is significant, so that often the redundant information is determined dynamically, and the pattern entries are embodied with the document as they are found.

.

Lossless Compression

The principle for lossless compression is to examine the bit-patterns of the data for repeated patterns. [[to document. page, word, related to indexing
example @ Gtext, 750Mwords 900,000 unique, compresible by word based Huffman coding to 20MB , but 10.5 dict, but unique words can be omitted 5M --> prefix tree 1M ['canonical huffman coding] Index compressed to 6% by delta coding bitmap (partially due doc.granularity/ else about 15-20%)

[[use material from CS545I lecture]]
Examples of lossless compression are the Graphic Image Format (GIF) (8-bit color) often used in Web pages, and the BitMaP (BMP) (24-bit color) used initially by Microsoft Windows and IBM OS/2. In GIF files each 8-bit byte points to a palette table of 256 colors. That palette, or a reference to some standard palette, is transmitted with each image.

Other techniques such as wavelets and fractals,/em> are also being incorporated into compression techniques along with a variety of error detection and correction techniques.

Lossy Compression

[[use material from CS545I lecture]]
Examples of lossy compression are the image JPEG format established by the Joint Photographic Experts Group, also used in Web pages, and the video MPEG formats established by the Motion Pictures Experts Group.

Lossy compression disables effectively some protection techniques, as digital

ENTEDU.Technology.simulation

Financing – from the economist

Who pays to study?

Jan 22nd 2004
From The Economist print edition

When universities depend on taxpayers, their independence and standards suffer

IT IS depressing to visit Oxford or Cambridge these days. The old buildings are so wonderfully grand that they highlight the cheap, ugly and badly kept new ones. The intellectual history is stunning, too: this is where Newton pondered gravity, and Occam honed his razor. But these days academics at Britain's two finest universities are a harried, ill-paid lot; salaries start at a mere £14,139 ($25,733).

Few disagree that both universities are living off the past, in everything from cash to reputation. The colleges' wine cellars are better than the kitchens, quips one don. The port and claret were laid down in happier times, when cash was flush and planning for the future mattered. But the food that goes with them is often dismal: that must be bought out of current income, which is usually earmarked already for everything from maintaining ancient buildings to supplementing salaries.

Reforming university finance
Jan 22nd 2004

Higher education

Britain’s government publishes its proposals to reform higher education. The EU gives details of its study on the world’s universities. The International University Bremen is a private university in Germany. See also the education ministries of France (in French) and Germany.

Yet Oxford and Cambridge are still in relatively good shape, thanks largely to their structure of self-governing, self-financing colleges. This limits the power of bureaucrats, provides independently managed money and ensures some protection for the original and the excellent. Other British universities have much worse problems.

To begin with, they have little or no endowment income to fall back on. The combined investments of Oxford and Cambridge are £4 billion; the rest of the British university system has £1.7 billion to play with. In America, Harvard alone has twice Britain's total. The “funding gap”—the hole in the universities' collective accounts created by the unfunded expansion of the past 20 years—is around £10 billion.

It is not just that money is short. The price and quantity of courses are state-controlled, in a system more suited to Soviet central planning than to a modern democracy. And as with other planned economies, the result of government intervention is increasingly unsatisfactory. In Britain, over 30 years, universities have gone from being almost wholly autonomous, with state-financed block grants handed out at arm's length, to becoming branch offices of a government ministry.

Admissions, too, bring a whiff of the old Soviet system. The government is convinced that more working-class students, including many with few formal qualifications, should go to university. Its ultimate target is 50% of 18-30-year-olds by 2010, and it is getting there fast. Figures released this week show that the number of students in higher education has risen in just one year from 43% to nearly 45% of the relevant age cohort. In 1979, the percentage of school-leavers going on to higher education was just 12.4%.

But more does not always mean better. One of Britain's best-known academic institutions, the School of Oriental and African Studies in London, found itself penalised for taking too few students from “non-traditional” (meaning poor) backgrounds. So it reduced entry requirements for such applicants, to take account of their often modest school results. But then it turned out that those students found learning Arabic or Chinese from scratch so hard that they were dropping out, incurring a further fine from the government.

The story of British higher education is less about expansion than inflation of qualifications. University degrees mean less and less and there are more and more of them. The rot set in in 1992, when the Conservative government allowed the polytechnics—locally based institutions that originally specialised in vocational teaching—to relabel themselves universities. That created a panoply of new academic courses, many of dubious merit, and kicked away a vital pillar of the higher education system, between the purely vocational further education colleges and the fully academic universities. This trend towards uniformity has disastrously weakened higher education in Britain.

Hence the importance of the government's proposed reform of university finance, which will allow a modest liberalisation of tuition fees. Instead of the current flat rate of £1,125, universities will be allowed to charge up to £3,000. The scheme is festooned with carrots, chiefly easy terms for poor students, in order to forestall a revolt by the government's nominal supporters in Parliament.

Critics say the new fees will create an unmanageable debt burden. Yet a broadly similar system in Australia has not had this effect: graduates pay back the loans when they are earning enough.

The scheme's real weakness, as most of the best universities admit in private, is that the top fee should be a lot higher; the cost of actually teaching an undergraduate is at least £10,000 in the humanities, more in engineering and science. But the most welcome ingredient is variability: universities will at last have the chance to offer cheaper, shorter courses to students willing to pay. The misguided notion that all courses at all universities are equally good seems about to be punctured.

Woes across the Channel

The present picture in Britain may be dismal, but misery is relative. Strolling happily through the Oxbridge quadrangles, and in the bustling corridors of less beautiful British universities, are 12,000 undergraduates from other European Union (EU) countries. Their home universities are in a still worse state: not only more overcrowded, but with barely a vestige of direct teaching. Oxford and Cambridge, more than other British universities, still offer undergraduate students close attention from a designated don.

The system is threadbare and arguably wasteful, especially as many students do little to prepare for their supervisions. But at least it happens. At France's best-known university, the Sorbonne, a translation seminar at the start of last term had 80 registered students. “Too many,” said the teacher superciliously. “Half of you have to leave. When we are down to 40 I'll start teaching. Foreigners will go first.”

In Germany, too, where professors enjoy the status of tenured civil servants, conditions are frequently dreadful. A current scandal is the Blockseminar—an ingenious system whereby an academic turns up briefly at the university and delivers an entire term's teaching in the space of a weekend, before returning to the unhurried pursuit of private knowledge.

Similar stories come from Spain and Italy, where universities are plagued by rigidity and corruption. Last year, students at Rome's Sapienza University were found to have paid up to €3,000 ($3,400) to pass their exams; and a professor at the University of Bari was arrested for demanding sexual favours in exchange for getting candidates onto the psychology course.

In effect, universities in these countries have become government-owned degree mills. Their aim is to get the greatest number of young people in and out for the least money and trouble. Really determined students may fight their way through to gain a professor's attention, win a research scholarship and start doing some real work, probably in postgraduate study. The others will arrive in the labour market, qualification in hand, feeling that their mostly middle-class parents have something to show for their taxes.

It is not all gloom and doom. Most countries have islands of excellence: German postgraduate engineering faculties, for example, or the French grandes écoles, fiercely competitive and independent. Finland and Holland have largely managed to keep quality up and bureaucracy down. But for the most part, universities in the larger countries of continental Europe are a dreadful warning of the consequences of nationalisation.

No wonder, then, that British and European academics cast envious and wondering eyes at the American university system. It manages both quantity and quality: more than 60% of American high school graduates at least start some form of tertiary education. And it keeps standards high, too. The European Commission recently published a painstaking ranking of the world's best universities, compiled by researchers in Shanghai. Of the top 50, all but 15 were American. From Europe, only Oxford and Cambridge made it into the top 10; from other EU countries, no university ranks higher than 40.

The American system is not flawless. The diversity which makes the system so dynamic also leaves it vulnerable to abuse. In the humanities, intellectual fashion seems bizarrely distant from the real world. Many bad ideas—notably political correctness—started life as American campus fads. And budget pressures squeeze the system when times are tough. This year, the axe has fallen hard on California's public universities.

Yet for all that, the numbers going into American higher education continue to rise, and the average tuition fee in an American university is around $4,500—some $1,000 less than the proposed maximum to be charged in England. Fees in the California state system, even after two steep recent rises compelled by leaner budgets, are less than $3,000, and a third of the income from them goes into grants for students who cannot afford even that.

Degrees of difference

Why does America succeed where Europe fails? The most important factor is diversity. American higher education is not just more varied, but has less of the crippling snobbery and resentment that accompanies variety in, say, Britain. At the bottom of the pyramid are community colleges, offering inexpensive, flexible, job-focused courses for millions of Americans each year. They are pretty basic, and Britons sniff at them. But the difference in mentality, says Martin Trow, an observer of both the British and American education systems, is that in America “something is seen as better than nothing”.

Crucially, too, the different bits of the system fit together. As Mr Trow points out, a student can start in a California community college, earn some credits, move on to state university and finish up taking a degree at Berkeley. Such a path would be inconceivable in most countries in Europe. In France, for example, the division between the state-funded, mass-market universities and the grandes écoles is vast and jealously guarded. Britain's further-education colleges are the poorest relations of an already impoverished family.

American universities are also fiercely competitive: for talented staff and students, for donations, for results (though competition on fees at the top end, where tuition can cost tens of thousands of dollars a year, is yet to come). Fund-raising efforts at the best-organised universities start even before students have graduated. Star professors attract star salaries.

That contrasts with the two extremes across the Atlantic. In Britain, performance is so minutely measured by the state that it stultifies the efforts of the brilliant, without really rooting out the incompetent and lazy. State supervision, coupled with penury, gives universities the smell of a failing nationalised industry, rather than of world-class outfits devoted to the risky business of thinking original thoughts.

In much of continental Europe, the problem is that senior university staff are not scrutinised enough. The intention, to keep academic freedom sacrosanct, is admirable, but the cocoon has become a prison. German academics are all but forbidden by law from getting involved in business. The best motivators for academic excellence are money, recognition and team spirit. But the German system penalises success in the name of equality: a university that does too well in the eyes of the federal bureaucracy will have its funding cut. So great is the risk of entrenched mediocrity that the chancellor, Gerhard Schröder, has urged the creation of—horrors—ten new elite universities.

A crucial part of competition is flexibility in setting fee income. Most European countries charge little or nothing. But fees have two beneficial effects. The first is that the university is beholden to nobody in its planning. Engineering and medicine are expensive to teach, so they cost more. Law is in high demand, so it is rationed by price at places like Harvard. But these are the university's own decisions. If it wants to teach something expensive, it can raise the money from fees, or from outside donors, or subsidise it from its endowment. It is not left, as Britain's academic managers are, wondering if it can squeeze money from the English department to keep the chemistry labs open.

Fees also mean that students are much more motivated. Underpriced goods and services are usually wasted, and university education is no exception. In a new book*, Robert Stevens, an academic who has run colleges in both America and Britain, writes of “an alcoholic yobbish culture” among students, for whom university is principally “a rite of passage”, like national service in the army, rather than an education. When Austria introduced a modest tuition fee of €363 per term in 2001, the number of students enrolled dropped by a fifth. Many, it seemed, were signing up simply for benefits such as health insurance.

But fees will also make students more powerful customers. Teaching at American universities is much better presented than in most European ones. Visiting American students are often startled to attend lectures with no visual aids, out-of-date hand-outs and droning, inaudible speakers. Such complacency will not long survive when customers have a choice.

The last big issue is selection. In most of continental Europe, this is a taboo. Access is either entirely open to anyone who has passed the school-leaving exam, or, at most, is rationed according to the marks gained. Universities, in effect, have to take the students the government sends them.

That sounds good, but works badly. The advantage of university-based admissions is that academics end up choosing the people they really want to teach. Students are more likely to focus on the course they want to study, and to try to meet the university's specific requirements.

Dream on, spires

American universities, with their mighty reserves of talent and money, look well placed to compete with the world's new academic powerhouses in India and China (which last year alone produced 2m graduates). How can sleepy Europe and timid Britain even hope to keep up?

The best hopes are in the piecemeal changes that are already happening. Students, for example, are voting with their feet. Britain's Open University, which offers part-time courses by post and e-mail, says that young people of university age are its fastest-growing bunch of students, up nearly 5% this year. That suggests that the disadvantages of a dumbed-down full-time undergraduate course, with the attendant debts and time spent not earning, are beginning to bite.

Employers too are signalling that there are too many graduates with indifferent qualifications. With luck, the British government's ill-starred 50% target may turn from its original force-feeding of the universities to a harmless exhortation that people should do something educational at some point after they leave school.

The days of social engineering may also be drawing to a close. The British government's proposed “access regulator”, an official body originally designed to force the top universities to take fewer students from fee-paying schools and more from poor backgrounds, seems unlikely now to penalise anyone. Just as well. Harvard and Stanford are both shopping for talent at Britain's top private schools, where pupils have been deterred from studying in Britain by official contempt for their class.

New institutions have sprung up, too. In Germany, the city-state of Bremen has set up an independent private university in conjunction with Rice University of Texas. “We wanted to be able to select students, to charge tuition fees, to have excellent and competent professors, to teach in small groups and in decent working conditions,” says Fritz Schaumann, its director.

Five years after its foundation, the International University of Bremen has 500 students, who contribute just over €3.5m in fees. It raises a further €20m a year from endowment income and donations. Other German universities at first regarded the newcomer with great suspicion. Now they are co-operating, for example in joint research programmes. Eventually, says Mr Schaumann, they will have to adopt a similar model.

Old institutions are also behaving in new ways. Britain's London School of Economics (LSE), for example, has largely escaped from the state's clutches. It now gains most of its income by selling courses to students from outside the EU, whom it can charge market fees. With that money, it can afford to hire world-class staff. “This is the only way we can compete with American academic salaries,” says Sir Howard Davies, the LSE's director.

For Britain's best universities, the big question now is whether to wait for more denationalisation, or to move towards freedom on their own initiative. For Europe's universities, the question is whether they can stop talking about reform and actually introduce some. Meanwhile, America's universities, hugely wealthier, happier and brainier, march remorselessly on.

*“University to Uni: The politics of higher education in England since 1944”. Robert Stevens, Politico's, £15.99