Difference between revisions of "Talk:Lecture 1"

Revision as of 22:03, 2 October 2004

Lecture 1 Discussion

Welcome to the Discussion Page for Lecture 1. Please use the + sign in the top of the screen to add comments to the page.

Article on 'hijacking' of cumulative innovation

Interesting article from First Monday:

The economics of open source hijacking and the declining quality of digital information resources: A case for copyleft

-- Tapan

Additional Readings

The following readings contain additional information on last night's material:

1) On Whether Patents are Good for Software Development:

James Bessen & Eric Maskin, Sequential Innovation, Patents, and Innovation, (Jan. 2000), available at http://www.researchoninnovation.org/patent.pdf

2) On the Liberality vel non of IBM’s Licensing Policy:

Ronald Mann, The Myth of the Software Patent Thicket (2004), available at http://papers.ssrn.com/sol3/papers.cfm?abstract_id=510103

3) Prof. Scotchmer's forthcoming book: Suzanne Scotchmer, Innovation & Incentives (MIT Jan. 2005). (The book is available at http://mitpress.mit.edu/catalog/item/default.asp?sid=872E36D5-2D83-4889-B3C4-DDB4DEF36AE3&ttype=2&tid=10277)

4) Finally, just for fun, two Golden Oldies. As far as I know, nobody has noticed them before. The first (I claim) shows that the key Internet technology of packet switching already existed in the Soviet Union ca. 1948. The second presents an anecdote describing IBM's experiences with "the anticommons" during the 1950s.

A Great Novelist Foresees the Internet

"Clipping, damping, amplitude compression, electronic differentiation and integration of normal human speech were engineering desecrations comparable to the dismemberment of a southern resort area like Novy Afon or Guzuf into little fragments of matter, stuffing them into a billion matchboxes, mixing them all up, flying them to Nerchinsk, sorting them out and reassembling them in their new location so that the result could not be distinguished from the original -- a recreation of the subtropics, the sound of waves on the shore, the southern air and moonlight." A I Solzhenitsyn, The First Circle (Harper & Row: NY: 1968). T. P. Whitney, Trans.

The Anticommons at IBM:

“At an August 1955 meeting with key engineers in Poughkeepsie, Watson Jr., chided them from hesitating to use ferrite cores in commercial products because of Wang’s patents. He quoted his father as saying, ‘That’s the most ridiculous reason for not moving into a new area that I’ve ever heard of because, one way or another, we can negotiate with Wang.’

“Indeed, the company was able to obtain rights to Wang’s patent for the reasonable fee of several hundred thousand dollars. . .

“Watson’s confidence that rights to any patent could be obtained at a reasonable fee, or the patent circumvented, appeared to be justified. But by the end of the decade, his confidence was deeply shaken. [Jay W.] Forrester had filed a a patent application under which MIT (through its patent management firm, Research Corporation) was demanding royalties of 2 cents for every magnetic core used in a coincident-current memory. This demand, made in November 1959, was quickly rejected for IBM by James Birkenstock who noted that ‘in our core storage units we employ seven of our own patents, as well as having acquired licenses under five patents from outsiders which were necessary to make the Forrester patent usable.’ If a royalty of 2 cents per bit were demanded under each of these patents, the cost per bit for royalties alone would be 26 cents, making core storage economically infeasible. Based on this analysis, Birkenstock concluded that a royalty of 2 cents per bit was ten to twenty times too much. But Research Corporation indicated it had already rejected an offer of 1 cent per bit, so an impasse resulted.”

[In an October 1960 interference proceeding, Forrester lost its ten broadest claims to a rival patent held by RCA. IBM had a cross-license to RCA’s patents.] “The jubilation was short-lived, however, as Research Corporation continued to demand royalties of 2 cents per core and initiated a civil action to regain the lost patent claims. Thus while IBM engineers were developing cost-effective memory designs and improving manufacturing techniques, litigation between RCA and MIT plodded slowly in the courts toward a settlement that would have profound ramifications on IBM and the entire data processing industry . . .

[IBM experimented with several other technologies, including “the Flute.”] “The experience with the Flute helped to emphaisize that there was no attractive alternative to coincident-current 3D selection ferrite-core memories. IBM had obtained rights to use all relevant patents except Forrester’s – and the validity of that patent was still being argued in the courts.

“Meanwhile, IBM engineers cointued to develop improved ferrite-core memory design and manufacturing methods as customers ordered more and larger memories. By the end of 1963, the company’s annual production of ferrite cores exceeded 1 billion; a royalty of 2 cetns per core would alone have cost more than $20 million per year. With the IBM System/360 slated for announcement in early 1964, the situation was intolerable. In February, 1964, an agreement was finally reached in which the company agreed to pay a on-time fee of $13 million for the use of Forrester’s patent if at least one of the claims was upheld in the litigation between MIT and RCA. The following month RCA and MIT reached an agreement in which the validity of Forrester’s patent was affirmed, and IBM made its payment to MIT. Larger than any previous payment it was nevertheless dramatically cheaper than the requested royalty of 2 cents per bit.”

C.J. Bashe, L. Johnson, J. Palmer & Emerson Pugh, IBM’s Early Computers (MIT: 1986), pp. 267-70.

Books Mentioned During Lecture

Two books were mentioned during the lecture by Ed Lazowska and Steve Maurer. One was http://www.amazon.com/exec/obidos/ASIN/087584863X/qid=1096603863/sr=ka-1/ref=pd_ka_1/102-7272045-1856915 Information Rules]. The other is http://mitpress.mit.edu/catalog/item/default.asp?sid=872E36D5-2D83-4889-B3C4-DDB4DEF36AE3&ttype=2&tid=10277

The two books complement one another quite nicely. Varian & Shapiro is already a classic. It tells a lot of stories and helps build intuition about what's going on in the new economy. The approach is in the general tradition of "how-to" business books. Prof. Scotchmer's book is just coming out from MIT next January and can be ordered through their catalog. It follows the economics tradition by starting with a handful of assumptions and working out the consequences through simple models. (Full disclosure, I'm a co-author on 3 of the chapters -- smm) It helps you learn to think about IT problems more abstractly. This is obviously a good approach for policymakers, but I also find that the approach provides nice insights for lawyers or businesspeople trying to design practical transactions. I'd say the technical level is a bit easier than the Procuring Knowledge reading, but the main point is that non-social scientists shouldn't be intimidated. Partly, that's because the book is modular -- i.e., it was deliberately written so that you can skip the formal proofs and not lose the thread. Second, I think it's a good general principle to get in the habit of reading books that are slightly too hard for you. That's equally true for IT people reading social science and for social scientists reading about technology! BTW, social scientists use the same math as the sciences, so many tech folk will find the subject surprisingly transparent even at a formal level.

Discussion of P=MC and entertainment media

From: csep590tu-admin@cs.washington.edu [1] On Behalf Of Ed Lazowska Sent: Saturday, October 02, 2004 2:43 PM To: Stephen Mark Maurer; isking@u.washington.edu Cc: csep590tu - Mailing List Subject: RE: [CSE P 590TU] P=MC and entertainment media

Just by the way, there is a service that has been rolled out in a few cities that nicely complements NetFlix.

You get something that's pretty much like a Tivo box. It comes pre-loaded with the 100 most recent DVD releases. Using a sub-carrier of the local PBS television station, the company trickles down new DVD releases to the box, displacing older ones -- it's a pure FIFO strategy, so the box always has the 100 most recent releases. When you watch a video, the box logs it, and once a week it calls Master Control and reports your usage; you get a monthly bill. There are only about 30 new releases a month, so the bandwidth requirements are minimal.

The technology comes from a Seattle company; I believe it's Disney that's rolling it out; it's in a few cities at present (but not Seattle).

The cost to "do a city" is only about $100K -- the cost of adapting the TV station to transmit the new releases on the sub-carrier.

VERY clever. Cheap as dirt, and it replaces the "front half" of your neighborhood video store -- the new releases. With no "Crap! All 16 copies have been checked out" problem. NetFlix, as Ian notes, has already wiped out the back half of the store.

From: csep590tu-admin@cs.washington.edu [2] On Behalf Of Stephen Mark Maurer Sent: Saturday, October 02, 2004 9:41 AM To: isking@u.washington.edu Cc: csep590tu - Mailing List Subject: Re: [CSE P 590TU] P=MC and entertainment media

Dear All,

Ian, thanks for this. BTW, lecture-specific comments should probably go on the Wiki rather than e-mail, particularly when they're nice enough to keep the way this one is.

It sounds like what's happening here is that the bricks and mortar world has MC = Price of Downtown Real Estate = High, and the streaming world will eventually have MC = 0. But in the meanwhile, we have Netflix which still has to pay for real estate in the form of a warehouse in a disreputable part of town, MC = Low, but also gets a significant amount of free space from the fact that its inventory is sitting in the mail and/or people's houses. One interesing question is whether Digital Rights Management (DRM) would be useful here. Lots of businesses give away CDs like a disposable item. Suppose I got the CD but had to telephone Netflix with my credit card number each time I wanted to use it...

So yes, you make a nice point. Here, DWL has a cultural face: Instead of everybody watching the same stuff, people with niche interests can look at their first, best choices. Kind of like what happened between the days when everybody read TIME and NEWSWEEK and the birth of the Web.

Have a good weekend!

smm

> All, > > There's a good article in this month's (October 2004) Wired magazine > called "The Long Tail", that is a corollary of the discussion we were > having about economics when the marginal cost is zero. Basically, a > bricks-and-mortar outfit such as Blockbuster (to focus on video > rental) can only carry so many (roughly 3,000) titles because of > economic limitations that drive physical limitations (i.e. a > Blockbuster store acres in size would dramatically increase MC through

> overhead); therefore, they carry those titles they believe have the > broadest popularity, to maximize return on their MC (which is cost per

> unit space of real estate to display a DVD). However, a service such > as Netflix can afford to carry titles that may only see 100 rentals in

> a month, or a year - but they can carry a LOT of them, and over half > their business is outside this "major market" catalog (thus, the "long

> tail" of low popularity content). They can afford this because their > MC is lower than that of Blockbuster; if they rent "Forbidden Planet" > once a year, it's revenue above and beyond their overall MC per title > (or they're being stupid and won't be around much longer). Other > forms of media (e.g. streamed/downloaded music) see the same benefit -

> and the article mentions in passing the availability of retro game > software within this same model. > > The moral: if it doesn't cost you anything (extra) to deliver a > product only a minority want, you're still making money on that > minority. If you don't serve that minority, they're your dead weight > loss. -- Ian >