IP: more on On State Universities, S-2031, and Bayh-Dole

[Dave Farber <dave () farber net>]

------ Forwarded Message
From: "Jonathan S. Shapiro" <shap () eros-os org>
Date: 08 Apr 2002 17:45:24 -0400
To: farber () cis upenn edu
Cc: darren.lacey () jhu edu
Subject: Re: IP: On State Universities, S-2031, and Bayh-Dole

Dave:

I'ld like to offer a small correction to the previous note concerning
Bayh-Dole and share my own impressions as someone who is dealing with
the consequences of the law.

First, it is telling that Seth Johnson's note doesn't bother to give a
URL for the law itself. Johnson is flat wrong on some important points
and he doesn't really want his readers to go read the law to find that
out. Bayh-Dole is officially known as "37 CFR 401". The text of the law
can be found at:

    http://www.iedison.gov/37cfr401.html

A reasonably clear and non-partisan description of the origins of the
law and the current state of the regulations was created by the Council
on Government Relations (a university-driven group). Their note can be
found at:

    http://www.cogr.edu/bayh-dole.htm

There is a nice set of bullets about midway down explaining the law in
English.

In summary: Bayh Dole was enacted because the government wasn't
generating value from its patents. If you think *universities* are bad
at patent management, imagine how bad the government is at it. The
government goal was to increase technology transfer. They clearly
succeeded. It is not clear whether Bayh-Dole was the best way to do it.
Some of the evidence gathered by policy researchers that its effects
have been bad both for technology and for universities.


Contrary to what Seth Johnson says, Bayh-Dole does NOT give the industry
exclusive licensing rights to Federally funded research. The act does
three things:

  1. It says that Universities may *elect* to retain title
     (i.e. to own) the patent rights on government-funded inventions.
  2. It requires the universities to make an explicit
     decision: if the universities choose to own the rights,
     then they *must* patent.
  3. If the university fails to patent, the government agency
     *may* patent in the name of the United States.

Item (3), in practice, almost always results in the technology ending up
in the public domain.

While a University is free to set any licensing terms it wants on
patents -- including issuing exclusive licenses -- it is not required to
do exclusive licensing. Much depends on the specific university. The
catch lies in Item (2) -- since the universities don't have a budget for
patent filing they tend to get this funding from outside parties, which
tends to lead them to make exclusive licensing deals. While they *can*
walk away, most universities don't. They view the filing as an
opportunity to pull in money from licensees. This is purely a numbers
game -- more patents is better and the system tends not to examine the
relative value of the patents.



In practice, the Bayh-Dole act encourages Universities to be
greedy-stupid. There are two ways to set up a technology transfer
office:

  1. Office must be self-funding.
  2. Office is underwritten by university with goal of long-term
     returns

Most universities, including Johns Hopkins, have gone the first route.
Instead of looking for long-term returns on research, a self-funding
technology transfer office seeks immediate payout. When a company wants
to license the patent, the self-funding office looks for a flat fee and
possibly a per-copy royalty stream. Generally, the license is exclusive,
because if it isn't exclusive it has no value (Universities, in
practice, do not enforce patents). The catch? The early value of the
patent is generally pretty low -- it is an unproven technology. A patent
that might eventually be worth millions goes for a one time charge that
is generally tens of thousands of dollars.

The self-funding technology transfer office has a strong disincentive to
patent: filing a patent costs money, and the self-funding office doesn't
have any. This tends to lead to a situation in which the licensing
company is effectively paying for the filing, which tends to reinforce
the short-term world view of these offices. In order to get the benefit
of the addition to their operating budget, the University will often
license patents at very low prices on a flat fee basis -- giving up
potentially massive returns in the process. As I say above, this is a
numbers game: more patents means more money. Quantity, rather than
quality, becomes a goal.

A big problem here is that this approach balkanizes the intellectual
property. Once a technology is exclusively licensed, it cannot be
licensed again. Suppose my colleague comes up with something clever and
the university licenses it out. Two years later I come up with some new
application and we decide we want to do something with it. We can't,
because the previous license was exclusive. This is quite bad.


A few universities, including MIT (though MIT has gone back and forth on
this) take the long view: they will often license a patent in return for
an equity stake in a new company, hoping that the company will pay off.
These deals are usually a trade-off: the new company gets a time-limited
exclusivity with an option to extend or some kind of a flow-through deal
on third-party patent revenues. The basic idea here is that the
University is making an investment in the company (the investment is the
cost of the patent filing), but builds in some means for recovery. The
reason to play is that the university is, in effect, getting "founder
shares" that have potentially large pay-off. Further, by avoiding purely
exclusive licenses, the university retains the ability to spin out
companies that combine the work of several people.

The long-term technology transfer office requires a team that is very
well educated and can make good decisions about which patents to file.
This is very very hard to do, so patenting by the University doesn't
happen unless some faculty member wants to pursue an invention. While
equity stakes can be done, the university is invariably a minority
shareholder, which leads to SEC complications for the startup company.
At the end of the day this becomes so complicated that many universities
simply let their faculty run with things at no charge and rely on
donations back to the University endowment -- no contract, just an
expectation of reasonable behavior. It works surprisingly well for MIT.
Why? The professors gain (in their academic roles) from those donations.


The results for the University: MIT and other long-view tech transfer
groups do very well. Short-view groups do really badly. Hopkins, for
example, has made essentially no income for the Engineering school from
patents (the medical school has a long-view tech transfer group, and
does quite well). There is talk of closing the current technology
transfer office and starting over. I don't know how serious that is, but
the revenue results from the office have not been very good.


As a current university faculty member, my experience with Bayh-Dole has
been mixed. In practice, the person most likely to take University
research and transition it into industry successfully is the principal
investigator (the faculty member) who creates it -- other scenarios are
possible and sometimes successful, but this is the most common one. As
an inventor, I find that I must now pay my University for a license to
technology that I created that all of you (the taxpayers) paid for. This
is somewhat ridiculous. Further, I am competing with other companies for
the right to use my own work.

Fortunately, I am dealing with a technology transfer office that is very
well intentioned. It is also fairly young, and doesn't have a lot of
experience with software startups. Some of what the tech transfer office
wants to do fits the pattern of other businesses, but isn't very
workable for a software company. This is mostly an education problem,
and we are working through the issues.

The one thing operating in my favor is that software licensing is so
bloody complicated. There are no general guides and one deal does not
tend to create a good precedent for the next one. This leads to a
situation where I have a lot of input into how the deal is getting
structured. At the end of the day the most frustrating thing to me
personally is that the University (and also my lab) will lose if the
tech transfer office follows the self-funding approach of one time fees.

In fact, I find myself frequently taking the University's side in order
to get a better deal to happen. Hopkins, as a rule, is more concerned
with patent liability than with income. If I build a strong company, I
want the university (and my department and lab) to benefit. Hopkins has
a self-funded tech transfer office, so this goes against the grain. At
the end of the day, I may be better able to achieve benefit for the
university by not patenting anything and simply reserving an equity
stake (or some comparable arrangement) for the university in order to
bypass the short-term influence of the self-funding arrangement. We will
have to see how it goes.

In software, at least, the results for the country would probably be
better if nothing was patentable. Software companies don't need patents
to succeed. The key to success in a software company is visible,
continuous leadership on a technology.


For some faculty members the entire Bayh-Dole arrangement creates
tremendous anger. A few take steps to subvert the process intentionally.
One favorite is to publish in order to prevent patenting. This can
happen for several reasons:

1. The inventor may feel that the taxpayer (including themselves) is
getting ripped off.
2. The inventor may see future applications, and may want to prevent the
technology from getting trapped in a single licensee. Publication
becomes a way to trump bad decision making in the tech transfer office.
3. The inventor may want to use it themselves.
4. The inventor may have built on prior work that they own, and may not
wish to give the university exclusive rights over a derivative work.


As a taxpayer, you have already paid for this work, and you are being
deprived the use of the results. If I came to your door and said to you
"Here is the deal: you contribute $5 and I'll give the rights to the
results to somebody else", would you take the deal? Advocates of the law
will tell you "if investors cannot get exclusivity, they won't invest in
the next step of the technology development." This is flat wrong.
Investors invest in lots of things that are not patentable at all! They
do examine the investments more closely when exclusivity is not
available. This is probably a *good* thing.


At the end of the day, one part of the Bayh-Dole outcome *is* clear.
Universities are not in the intellectual property management business.
They are in the education and basic research business. Basic research
does not lead directly to products. Over the long haul, the effect of
Bayh-Dole is to encourage universities to foster short term (and
therefore more readily commercialized) research. This creates a basic
conflict of commitment, and it conflicts with the open atmosphere that
has made our university system successful. It has worked to date because
universities have a tradition of openness and academic traditions take a
long time to change.

However, if you want to understand "how will that institution/person
behave in the future", you need only ask: "what are the incentives?"
People and institutions get it wrong, but taken overall they try to act
in ways that are good for them (this is why policy has impact). Over
time, the effect of Bayh-Dole will be to move universities out of the
research business and into the technology business.

On the whole, this is bad for the country and the world. One by one we
have lost all of the institutions that created the basic science of the
20th century. Some of that basic research had noticeable impact:
telephony (Bell), economics (Nash, Princeton), electronics (various),
medicine (various). Universities will be the last to go, but if
Bayh-Dole and similar policies remain in force the universities will get
out of the basic research business.

Which leaves us to ask: who will do the basic science that will drive
the innovations of the next century?



Jonathan S. Shapiro
Assistant Professor, Department of Computer Science
Johns Hopkins University



------ End of Forwarded Message

For archives see:
http://www.interesting-people.org/archives/interesting-people/