Innovation Science

iPod Listening Safety

There is considerable interest in the amount of damage to one’s hearing that listening to an iPod (or similar portable audio player) can cause. I’ve posted before on Apple’s attempt to prevent harm to hearing from iPods and discussed possible improvements. In the absence of such a solution, it’s important that people understand safe listening levels for iPods and similar devices.

At the recent annual conference of the American Auditory Society, researchers from the University of Colorado and the Children’s Hospital Boston presented the latest data on how long one can safely listen to digital portable music players (Portnuff and Fligor, Output Levels of Portable Music Players).

Firstly, they showed that the danger among all of the players tested (iPod, iPod Mini, iPod Nano, Creative Zen Micro, Sandisk Sansa) are approximately the same. In other words, iPods are no worse for your hearing than any of their competitors.

Secondly, there was no difference in danger to your hearing between different music genres. Believe it or not, R&B music is as potentially damaging as Rock music or Country music. No word on the danger from Opera (although some might suggest that the true danger lies in falling asleep and being exposed to hours of Wagner—or perhaps the danger is in staying awake).

Thirdly, how long you can safely listen to your music player depends on what you are listening with. People choose different earphones to listen to their player: some use the buds provided with the player, some upgrade to expensive insert earphones such as the Etymotic ER6s or Shure E4cs. Some choose to use large headphones that sit over the ears. Depending on which you use, your safe duration of listening is different.

The table below shows the levels calculated by the Boston University researchers that reach the 50% noise dose per day according to NIOSH standards. Exceeding these levels is not a good idea.

For example, if you use Etymotic ER6s (categorized as “Isolators” in the table below) and you are listening with the music player’s volume control at 80%, then you should not listen for more than 50 minutes a day. The authors of this research report even suggest that “more conservative recommendations may be warranted.” Listener beware.

The View from Intel Research Berkeley

Last week I went to an open house at Intel Research Berkeley, a research center that occupies the penthouse of the 14–story building that also houses the research facility I run. The views of the Bay Area from up there are probably the best one can see from the East Bay, appropriate for the visionary and lofty goals of the projects that Intel showcased.

The mission statement of the research center is

Drive off-roadmap, high-impact exploratory research vital to Intel

a sentiment that should easily translate to research at most companies. Interestingly, the technology buzztheme Simplicity has made its way into the Intel Research vision:

Essential Computing: simplifying and enriching all aspects of work and daily life

The research center was created in a 2001 Intel initiative to reach out to University talent, opening similar centers near Carnegie Mellon University and University of Washington (Intel Research Berkeley is 1 block from the UC Berkeley campus). Surprisingly, according to Henry Chesbrough in his book Open Innovation, Intel didn’t have much of an internal research effort until 1989, with only external research supporting them until then.

Okay, back to Intel’s open house.

The diversity of the research presented was impressive. There were posters and demos for over 50 projects, categorized into the following broad fields:

• Security/Networking/Distributed Systems
• Human-Computer Interaction
• Technology for Emerging Regions
• Programming Languages

Each project was also assigned a descriptor for marketing appeal:

• Richly Communicative
• Concealing Complexity
• Personal Awareness
• Emergence Engineering
• Potpourri
• Physicality

The Security/Networking/Distributed Systems research included projects on power management for routers, fingerprint pattern recognition, and virtual machines for long-term data storage and retrieval. The details in most of these were fairly incomprehensible to me given my lack of expertise in these areas.

Of greater interest (and comprehension) to me were the Human-Computer Interaction projects.

One was titled Data Souvenirs: book-like objects that sat on your table or bookshelf and provided electronic information on the spine and inside the book, such as an lcd screen indicating when you have received e-mail from specific addresses, eliminating the need to constantly check your e-mail at home over the weekend/evening. One intriguing application of this project was called something like Real-Time Journey in which the book represents the long-term progress of an historical event in real time. For example, the book might track the voyage of Lewis&Clark, with the spine displaying their progress over the same time-scale as the actual trip (e.g., distance traveled or states reached). Inside the book would be maps and information on their voyage. As the representation of the virtual trip reaches a landmark (e.g., Mississippi reached), you could open the book and read about that point on their voyage. The possibilities for other creative uses are interesting, although the most difficult part of this project is probably the identification of the key value proposition to consumers.

There were several interesting cellphone projects, focusing on either an application or a combination of application and system architecture. One project addressed problems of application incompatibility between cellphones by designing applications that did not run on the phone itself. They demonstrated an application where people could make real-time adjustments to a remote display with their cellphone, the demo’ed application being a conference meeting where people are adjusting their responses in reply to questions from the speaker. As they made adjustments to their response with their cellphone, they could see those changes in real-time on the remote display being projected.

Another cellphone application was place-based ringtones, where the ring that your phone makes will depend on who is nearby and what signatures they have on their phone. In a separate cellphone alert project, a visual metaphor was demonstrated that would accompany ringtones: instead of simply flashing the name and phone number of the person calling, a landscape image on the phone display would change to indicate if the call was from someone you know, a specific person, etc. For example, a park scene might suddenly show birds flying in the distance of the call is long distance, or a swan floating on a lake if it was a friend. This would personalize the information and encode the data in metaphor that only you could understand.

One interesting project was a combination of human behavior research with statistical analysis for the optimization of computer power management. Several gigabytes of data were obtained from thousands of hours of pc usage to determine when is the best time to power down such computer resources as a computer screen or hard disk. The goal was to maximize battery life while minimizing human annoyance. The challenge was how to measure this passively, without constantly asking people what they like and don’t like. The researcher(s) ended up interpreting user annoyance by identifying associated behavior: if the user immediately shakes the mouse or mashes on the keyboard immediately after the screen powers off, that’s a sign that the person did not want the screen to power down at that time. Data on what the user had been doing and what states the computer was in before power-down was gathered to understand how to predict when the user does and does not want a power-down. A massive hidden Markov model was created from the 50–dimensional data set to predict human behavior such that the power management can better determine when to power components down, rather than rely on simple rules like “turn off the screen after 5 minutes of inactivity.”

There were several projects that focused on providing technology in emerging regions, such as West Africa and India. Applications ranged from providing WiMax connectivity and telemedicine in rural areas (see Cnet video), and literacy programs developed for cellphones. Several projects also addressed green issues.

I was impressed with the variety of projects at this one small facility. How they get that work done with those spectacular floor-to-ceiling views, I have no idea.

Responses to "How to Succeed with Industry-Sponsored University Research"

Sonicboom posted some great comments and questions in the Comment section of my recent post about industry-university collaboration. Because I have a lot to say in response, I’m posting my thoughts here rather than in the Comments section so that RSS subscribers don’t miss the discussion. You can read Sonicboom’s complete comment here.

Sonicboom asks:

What would you estimate is the percent of industry-sponsored research being performed at top-tier research universities?

The NSF reports that in 2004 $27.4 billion in university research was funded by the federal government while $2.1 billion was funded by industry. In the ‘02–’03 fiscal year, $30 million in industry funds were directed to University of California (UC) research projects through the UC Discovery Grant program, representing a fraction of the total industry-sponsored research funds that the UC system received that year.

The benefit flows both ways. University research has had a big impact on the biotech industry, for example. One-third of the public biotech companies in California were founded by UC faculty, including Amgen and Genentech. From 1996–2000, there were 172 UC research projects sponsored by the biotech industry alone, and the number of sponsored projects per year has certainly increased since then.

In your opinion, how efficient is the marketplace for turning ideas in university labs into marketable ideas? While you point out this is not the goal of university research, it is becoming a focus within most research universities with the proliferation of "technology offices" and incubators.

Transferring university research discoveries to the public domain so that the general public benefits from the results is not an efficient process in universities. The whole university research process has evolved such that research advances knowledge of a field and the results are disseminated through publications. There has been little focus on applying that knowledge or transferring it to industry in ways that benefit the general population.

Much of the engineering and scientific research at universities is motivated and directed towards real-world problems, but not all of it actually becomes used in a way consistent with that justification. For example, many of the speech, hearing and audio signal processing research grant proposals motivate their proposed research by saying that the results will help produce better hearing aids for the millions of hearing impaired people in the US. The reality is, though, that little of this research has had an impact on that industry.

The National Institute for Health is promoting a Translational Research initiative to attempt to transfer basic scientific results into the clinic where patients can experience benefit from the research. The NIH has motivated this by noting that the general public is funding the research through their taxes so they should benefit from this investment. Industry-relations and tech transfer offices at universities have some of the same motivations, in addition to seeking ways of generating revenue and research dollars outside of traditional avenues. As I pointed out in a previous post, however, there are several stumbling blocks to transferring university research to industry. They include:

1. a lack of knowledge by university researchers about industry state-of-the-art and technology needs necessary to refine the research in a way that optimizes its potential for application;
2. an attitude that applied research is distasteful and that working towards solutions for the general public is a violation of the ideals of scientific research (Going to the Dark Side is a term often used for university researchers who start to work with industry or for those who leave academia for a job in industry);
3. the need to openly publish research results that is often at odds with the competitive goals of a company;
4. the absence of someone who can find the sweet-spot in collaborative research who can define a project that will meet the needs of both the university and sponsoring company. This person would need to both understand the specialized university research and determine how the research could by applied to a product given the current state-of-the-art and the needs of the customer. This is a difficult task.

I hear bureaucrats drone on rather than creative, business-minded folks looking for ways to match the wonderful ideas they have in their labs with the companies that can turn those ideas into useful products.

As I just mentioned, matching university research to products is a great challenge. There are certainly many creative people at universities who could to do this, but they need a new type of training that typically doesn’t exist in universities except at places like Stanford, which has several programs for applying research to product development. The business people at universities facilitate the collaboration, but the definition of the collaboration requires specialized knowledge and exploratory work.

Additionally, research by Di Gregorio and Shane at the University of Maryland suggests that intellectual excellence and university policies are the most significant factors contributing to the creation of startups from university research, possibly accounting for differences across universities in startup activity. In 1997, for example, Stanford produced 25 new startups while Duke produced none, even though they both received from $300–400 million in sponsored research funding.

technology offices at universities…want to turn their  professors into entrepreneurs…This seems to conflict with the university’s primary goal of teaching…(and) to pursue their research interests.

Right, most professors aren’t interested in this type of work. A few universities like Stanford and MIT have successfully developed entrepreneurial programs over decades where the commercialization of research is accepted by faculty, and those universities now attract researchers who are looking to participate in this challenge. I believe developing such a system is not easy and requires a change in the mindset of the faculty and administration as well as the-development of appropriate policies and programs.

At it's core, the commercialization of university research seems like a search or matchmaker problem…This "search" problem, which I like to call it, seems to present an enormous business opportunity for those who can create the right model for commercialization.

I think you’re right. This relates to my point #4 above about identifying the sweet spot. There is an opportunity to be the bridge between universities and industry for those interested in doing the work to connect research with industry. The challenge is having the expert knowledge to assess difficult-to-understand research and envisioning the commercial opportunity which requires expert knowledge of the industry.

How to Succeed with Industry-Sponsored University Research

Companies often fund university researchers to advance new ideas in their field, and university professors often seek funding from companies who create products that are related to their research. A while ago I posted some advice on how to facilitate research collaborations between academia and industry. I promised to provide more information, so here it is.

Listed below are issues that the university researcher and industry sponsor need to address early in the discussion process. Often, both sides are unaware that these can be potentially contentious issues until well into the work and the issue is staring at them in the face, at which point the relationship can start to strain. Answer the following questions up front, and the collaboration or sponsorship will go much smoother.

1. Why is the company interested in sponsoring this research?
Different companies have different motivations for funding university research. Some simply want an association with a nearby university or a leader in their field for PR purposes. This looks good to their customers, and having the researcher mention the company as a sponsor at talks and in publications can enhance the company’s image. If this is the case, then there will be little pressure from the company to provide details on progress and few inquiries from the company’s researchers about the work.

If, however, the company is funding the research because they hope that the outcome will be useful for the company’s own research (a more likely case), then company researchers may want to meet with the university researchers at various points to discuss protocols, time-lines, preliminary results, etc. The company is paying for the research, so of course they are eager to know the details about the work and the results. This can be annoying to the university researcher in part because industry and academia operate on different time scales. What appears to the company to be infrequent requests for updates will appear to the university researcher to be constant hounding.

2. What are the expectations and deliverables?
This relates to the previous question, but is more specific. What does the sponsoring company expect in the way of updates? Do they expect to be able to send their scientists to the university lab every month to talk about the project, or are quarterly updates by e-mail sufficient? Do they expect a written report at the end of the project with all of the data, or do they want all of the software code and detailed protocols so that they can replicate the results at the company labs? Discussing this up front will avoid frustrations later on.

3. What are both sides’ rights to publish and present?
Companies survive because they obtain advantages over their competition. Many of their advantages are proprietary and companies protect those advantages by either patenting them or keeping them secret. Unless the company is only looking for good PR from the project (see point #1), the business executives in a company would rather keep secret any useful results from the sponsored project for as long as possible.

This, of course, conflicts with the university researcher’s goal of publishing and presenting the research that they conduct. This ability is fundamental to the researcher’s job and should be undeniable by the company. The happy medium is to allow the company to file for patent protection on any inventions that arise from the project before the invention is made public in a presentation or submitted manuscript. This is typically achieved by providing the sponsoring company with copies of what will be presented or submitted for publication at least one month in advance to allow the company to review what will be publicly disclosed and to file a patent application if necessary.

On the other side, the company may want to publish and present the results of the research at meetings that are important to their business. They may or may not want the university researcher to participate in these presentations. Again, expectations should be discussed so that both sides are aware of each other’s goals.

4. Is the sponsor agreeable to the university’s Intellectual Property policy?
There is the possibility that a patentable invention will result from the sponsored research. These days, every American university has formalized their own policy on ownership rights of any inventions, with details on what the rights of the sponsoring company are to the invention. Areas include rights to use, produce and commercialize, exclusive and non-exclusive. Also to be considered are rights when the only inventors are university employees or when there is shared inventorship between university employees and company employees. Be aware of the policy and make sure everyone is comfortable with it before beginning work.

Often, companies only sponsor research that they believe will not produce any inventions in order to avoid any costly licensing fees. Corporate executives will also often bristle at University IP policies. I’ve known some companies that have pulled the plug on sponsored research after significant work had been done to set up the project because corporate lawyers did not like the IP policy, and other companies that have selected which university to work with largely based on which university had the loosest IP policy.

In my opinion, these are over-reactions. Companies incorrectly think that universities will be as cut-throat in negotiations as their competitors, and that the university will become a threat to the sponsoring company if all the IP isn’t tied up with the sponsoring company from the start. Universities are in the business of teaching and training students and conducting reputable research, not generating massive revenue from licensing technology. Yes, they do want a return from their employees inventions, but no, they are not going to pursue monetization of the invention in the same way that IBM or Intellectual Ventures would.

5. Will the research be conducted by the professor, a post-doc, or graduate student?
Research at university labs is usually not conducted by the professor who directs the lab but by students working with the professor. The students are there to learn and be trained.

Whereas industry researchers have significant experience and are efficient in making decisions and finding the right approach, student researchers will spend a certain amount of time exploring new ideas and learning new concepts. They will take considerably longer than a dedicated industry researcher because they are learning and being trained. If the professor will actively participate in the project, however, then progress will likely be faster.

Post-docs will also work faster because they have more experience. Presumably, the post-doc is not as focused on learning as a graduate student is and they are focused more on progress and producing publications. So, the sponsoring company should know who will be doing the primary research so that they will understand the speed with which it will progress and set their expectations accordingly.

6. Is the university researcher a collaborator or contractor?
Is the expectation by the sponsoring company that they are simply writing a check and that all work will be done by the university researcher, or does the sponsoring company want to be an active participate? If they want to participate, then at what level: helping to develop protocols, replicating research in their labs, providing equipment, data interpretation?

The approach that I try to take at my research center is to be a collaborator with the university researcher. For me, this ensures that the research has the greatest likelihood of being useful for my company’s needs, and eases the transfer of technology from the university lab to my lab. In the absence of participation, the likelihood exists that any results from the university lab will just sit on a shelf, waiting for someone to read the report.

7. Are there opportunities to enhance the funding?
Sometimes there are programs available that add funding dollars to sponsored projects. The University of California has a mechanism called the UC Discovery Grant that doubles sponsorship funding through a grant submission process: if approved, the state of California matches the funds of the sponsoring company. Money is always tight, and it’s worth investigating whether the university or state has funding mechanisms that can be used to enhance the sponsor’s funds.

How to Create an Industry-Academia Mashup

I was invited as a speaker and advisor to a summit at Purdue University two weeks ago. The meeting was created to discuss how cross-department acoustics researchers at the university can collaborate on new ideas, create a new interdisciplinary research center, and develop corporate partnerships. I was the industry-researcher representative, and I spent a considerable amount of time talking about how the different departments can collaborate under a single research concept.

What I didn’t get a chance to talk about was what university researchers need to understand to better collaborate with industry sponsors. So, over a series of posts I will discuss a few ideas on this topic.

Academia is increasingly turning to companies for research funding as federal funds become harder to get. The University of California has developed a matching-fund program that generates $50 million in corporate-sponsored research each year. In this post, I’ll discuss some of the issues that a university researcher must deal with in order to have a successful partnership with industry.

1. General Relativity
Time moves at a different pace on a university campus than on a corporate park: academia seemingly travels near the speed of light because what seems like a short duration to them can seem like a lifetime to industry. The university research process allows researchers to contemplate problems deeply and discover patterns of thought that had been previously undiscovered. A couple of months is nearly a blink of an eye to a university research program.

Time literally is money in industry, however, and a week lost in contemplative thought is a week lost in revenue. When R&D departments solve problems, they are not searching for the ultimate truth but for the best approach that can be developed given their time and resource constraints, knowing that there’s always time for improvements later. University researchers are driven by “Why and how?”; R&D engineers are driven by “Make it work.” Both sides of an industry-university partnership need to be aware of the relativity of their perspectives on the passage of time. Otherwise, “frequent updates” by university researchers might be viewed as abusive negligence by the company.

2. Training Day
Part of a university’s function is to educate students, both in classrooms and in laboratories. Research labs consist of students and post-docs doing most of the heavy lifting so that they can learn the research disciplines of their own field and become experts on their topic of investigation. This, of course, means that those doing industry-sponsored research are spending a lot of time simply learning their craft. Which takes time. Which means you need to consider point #1. Post-docs, though, usually are already experts in their field and are more effective time-wise: less time spent learning, quicker at conducting the research. Corporate sponsors need to understand who will be doing the sponsored research—graduate students or post-docs—and what that means for expected time-lines.

3. Publish or Perish
A researcher in academia is judged by their list of publications. Companies typically want to keep new ideas proprietary to obtain a competitive edge. You can see the potential for conflict. University researchers will want to, and should be able to, publish the research that is being funded—otherwise they won’t maintain their grants and possibly their positions. This issue should be addressed up front and guidelines provided to allow companies to apply for a patent before any inventions are revealed in publications or presentations. If a company doesn’t want anything published, they shouldn’t fund university research.

4. Know Your Client
University researchers focus on proving hypotheses and developing new theories—they typically don’t know much about the practical aspects of a potential corporate sponsor’s applications or their customers. Why should they care? Well, this presents a problem when looking for a common bond between industry and academia to justify a partnership. If you don’t know anything about dancing, don’t expect to be asked onto the dance floor too often.

The National Institute of Health is currently trying to promote a Translational Research initiative, defined as bringing results from the lone researcher into the clinic and to the patient, i.e., provide an application to basic research. This involves clinicians collaborating with basic scientists and each understanding the other’s world. A similar level of understanding is necessary between the academic researcher and the corporate sponsor. If the researcher doesn’t understand the application, their research is less likely to be of value to the sponsor because of the unlikeliness of it translating successfully.

Industry-types are often suspicious of the usefulness of academia to their business, and university researchers need to propose ideas that demonstrate an understanding of the issues facing the sponsor’s industry. University researchers looking to create partnerships with companies need to spend some time understanding problems that companies have with their technology and their customers’ needs—find the areas of opportunity where university research overlaps with industry need. This produces the highest likelihood of a company being interested in sponsoring research and and of that research ultimately being useful to the sponsor.

5. Ownership
The area of intellectual property is such a quagmire these days and such an important part of university-industry relations that I think I’ll give this topic its own post some time soon. Sorry.

New Sennheiser Research Center

Sennheiser, a world leader in transducer and headphone technology, just opened a research center in Palo Alto near Stanford University (and half a block from the beloved Fry’s Electronics store!). I went to their open house reception today, which had several executives from their German headquarters, a member of the Sennheiser family, the mayor of Palo Alto, and many of the most prominent audio researchers in the Bay Area.

The recent openings of satellite research centers the Bay Area is an indicator that technology companies are doing well and investing in the long term. Research is usually the first thing to get cut at the slightest hint of financial difficulties, and opening a research center in the Bay Area produces a significant hit on a company's net profit. Such research centers typically do not provide positive support to revenue for many years, so their formation requires long-term dedication. The formation of these centers is also indicative of the realization by companies that they need to work harder to compete on technology development.