CS73N Meeting 05 Notes: Net Neutrality, Patents, Medicine
By Gio Wiederhold, recreated 22 April 2006, updated 24, 25 April 2006. Updated by Avron Barr 22 April 2007; by Gio 16, 19, 20 April 2008.
Topics Covered briefly
18 April 2007
Next Thursday, April 24, is a guest presentation on “Writing for the Web” by our friends from Wired Moon. Don’t miss it!
There is no specific assignment, but you all should have a functioning web page, reachable from the wiki classlist, by next Friday.
Internet Neutrality
Kamil Dada presented his observations from the Thursday, 17 April 2008, FCC meeting. The trigger was Comcast and other ISPs disabling transmission from Bittorrent, because of `excess load' on the system. The dominant class sentiments were for disallowing any control over transmission by the ISPs. However, some users, specifically folk that download movies can use a disproportionate fraction of the Internet, here, and even in japan, where the broadband is wider and distances are less. Government controls are seen as unwise, since the Internet does not stop at country boundaries, although China controls the Internet within its domains. Economic measures to control excess usage could be small charges for transmission or receipt, a periodic, say monthly, limit on free use, or combinations.
Kamil's report was published in the Stanford daily of Friday, 18 April 2008 and more is here.
Patent Trolls
A public access research website at Stanford, providing access to a database with HIV information, supported by NIH grants, is being sued by ABL who hold the 988 patent granted in 2001, which covers research done and published at Stanford in 1976 and again in 1984
by Ted Shortliffe and others (see references below). The Stanford Office of Technology Licensing (OTL) signed an agreement with ABL which essentially recognizes their patent, and forces the Stanford website to credit ABL with the invention.
We discussed the issues arising from such misuse of patents. The patents are typically owned by `patent trolls', companies who file or purchase patents that have never been reduced to practice. This misuse is made possible by the laziness of the patent office when granting such patents, without any review of the subject literature outside of its own patent files, and leaving it to the lawyers to sort things out. Small companies and startups have to concede when threatened, because they are not in position to fight, neither financially nor because of loss of venture and other investments when their IP seems to be threatened. But Stanford has the intellectual resources to fight, and would not be threatened by loss of venture investment. If anything, Stanford will gain recognition in the Silicon valley and beyond as an organization who will stand up against such nonsense, which ruins innovative industries.
The class felt unanimously that Stanford should not roll over, but be a hero.
Abstract of patent (http://www.patentstorm.us/patents/6188988.html)
Shortliffe, E.H. Computer-Based Medical Consultations: MYCIN,
Elsevier/North Holland, New York, 1976.
Buchanan, B.G. and Shortliffe, E.H. (eds). Rule-Based Expert Systems: The MYCIN Experiments of the Stanford Heuristic Programming Project. Reading, MA: Addison-Wesley, 1984.
Medicine and the Internet
Computing has invaded medicine more than education. Stuedents felt positive of waht computing does now and will do for medicine. Billing depends on computer entry of patient data, diagnosis codes, and prescribed treatments. Medical records are being computerized. laboratory test results are transmitted on-line. Telemedicine enables care to be delivered to rural sites and remote countries, and patients will look up their symptoms before going to a physician. Diagnostic radiology is being outsourced to countries where experts are available and get paid (much) less. Compounds for potential treatments are assessed using computer analysis and statistics from patient trials. Genomic matching is starting to inform patients about propensities for diseases. Some metabolic pathways can be simulated in order to predict outcomes when the pathways are being affected by drugs or other interventions. Medical research results are available on line.
Problems do exist. The patient-provider interaction can be hindered by distraction [P. Hartzband and J. Groopman: Off the Record — Avoiding the Pitfalls of Going Electronic; The New England Journal of Medicine, 17 April 2008]. Poor record collection makes the on-line medical record less valuable. Prognostics are rarely entered, and on subsequent visits the provider, if they differ or don't rember, looks at prescribed medications to guess what the planed treatment was.
Suggestions for improvement include much more automation of information collection at the patient interview sites, unobtrusive screen, patient participation in the interaction, including a touchscreen for marking issues, and feedback about the effects of candidate treatments. Help in diagnosis, by combining patient information with collected, abstracted and processed historical information about similar patients in the past, and for predictions specific simulations for the patient's profile. Such services are challenges beyond waht computers have been able to produce today.
Date formats
We briefly discussed why we use a date format with letters. It is internationally unambiguous, and in the end ddMMMyyyy does not take more space than mm/dd/yyyy or dd.mm.yyyy or yyyy.mm.dd, other formats in use.
History
We looked at the basement history displays in the break Items were
- Storage devices, from 3 relays on a card per bit, to vacuum tubes, holding 1 bit, to transistors, to multiple transistors mounted together , to ever larger integrated circuits with many millions of bits. It takes 8 bits to make a byte, the common representation for a character.
- A photograph of Gio with some 1977 students and post docs. several are now professors in other places, some have started companies, as David Shaw, who started a hedge fund. Jerrold Kaplan started a tablet company, examples seen in the case, but was too early. He recovered by publishing a successful book though, Startup.
- The third case is devoted to Xerox PARC, a great research organization that was not able to switch to making and marketing products.
- The fourth case shows the first IBM PC, which came out in 1982, with a card for the alternative to Microsdoft Windows, the CP/M system.
- On the far side is the original 1998 Google storage, 10 diskdrives in a Lego case.
Projects in terms of the future:
The available technology will certainly be different.
There will ubiquitous mobility. There are many opportunities for sensors [smell, temperature, skin resistance, telepathy?} in addition to sound, vision, and electro-magnetic radiation.
The rate of change is affected by changes in the social fabric. Our personal interaction is changing rapidly, but our interaction with pets, as dogs is changing less.
Will simulated animals replace real ones? Yes, to some extent, when space and time is scarce.
If people don't remember real pets anymore, will there be simulated fantasy creatures that might be more attractive? They should probably present problem as well, to satisfy the social need to instruct and fix things.
Will simulated children replace having real children?
The economy will be more global. Will it be more balanced?
When will have sustained colonies on Mars? 50, 100, 200, 400, 800, 1600, 3200., 6400, 12800, ... years
Education
We discussed education earlier. The internet has changed much in the way material is distributed, but not changed the structure all that much. While it is possible for a professor to deliver a lecture to thousands of students, the personal contact to follow up also has to be provided, but perhaps by different people.
Facilitators may be employed at remote sites to help and assist, as has been found to be successful in the Stanford TV courses.
Those might become candidates for professors in the future, since they will develop insights that are lacking in recent PhDs when those become professors.
For training, as in the military, computer based methods are already useful. Training should be done when needed, not when there instructors available.
In medicine, we can now teach future physicians to look things up, rather than remember everything. Teaching Medicine has two components, developing a high level understanding of very complex relationships, and training, so that performing the right action becomes routine.
Digital Government
Government is another area where improvement is needed. We had talked about voting.
Could using simulators or computer based monitoring systems increase the efficiency and decrease the cost of operating the DMV?
We did not yet talk about rule-making, and important area for public participation.
From Facts to Knowledge
Information is the intermediary representation that comes between facts and knowledge.
Facts of the real world can be encoded into a data [[NoteRepresentation
representation]] and stored for analysts to use. Analysts can process
that data according their knowledge, making assumptions, and publish it
as information. Information can also be further
processed, creating information further removed from the source of the
data and subjected to more assumptions. Such processing includes
selection, summarization, integration, abstraction etc.
Stored information does little good until it is
transmitted to a reader. The technical definition [Shannon] is more
narrow: information should be novel to the reader, otherwise it is of
no import.
For our work, we place a further constraint on
information: it should be actionable. That means the reader, in a role
as decision maker, should be able to base an action on it, say buy a
piece of hardware.
There is information that amuses or educates, and
as such is worthwhile. But for a Business on the Internet the
information should be actionable, even while it increases your general
knowledge.
Actionable information leads to actions, and those
actions will change the facts in the world: after some time there is
one fewer pieces of hardware in the store, and one more in your hands.
That fact can be recorded, and becomes part of the data cycle.
Trust
Trust in information you obtain is an orthogonal
issue. Earlier we focused on trust as derived from authority
of the author. But if you can track the process back that generated the
information, and validate the facts that wee used, you are even better
off. But that means dealing with many assumptions. We discussed some in
class, especially metrics, as should safety of flying versus driving be
measured by passenger miles or passenger hours. It matters less what
the right way is, as long as the reader of the information is clear
about what was measured.
