The next generation of academic publishing

(Note: my academia-related posts are strongly colored by my experience studying computer engineering. Other fields will/may differ.)

Since leaving the publication treadmill of academia, I’ve spent a lot of time thinking about academic publishing, and what it could/should be.

The publishing process.

Similar to most large entities, the academic publishing and dissemination process is amazingly slow. Here was the process for me during grad school:

  1. Do research and write up a manuscript (anywhere from 1 month to years)
  2. Submit to a conference
  3. Wait 2-3 months for review
  4. Get a review and write rebuttals
  5. Wait a few weeks
  6. Get an acceptance or rejection. If it is a rejection, wait a few months until the next conferences and go back to step 2.
  7. If the paper is accepted, work on the camera-ready and submit a few weeks later.
  8. Wait another 3 months and then go to the conference. At this point, it is officially published put in print and is published.

The entire process is around 7 months. That is a long time.. and that’s if you are lucky and get an accept the first time around!

Many papers go back to step 2 several times, meaning that it can be years before a piece of research is actually disseminated into the world. The rejections can be fairly random too. I have a friend who got a paper rejected 5 times (or was it more?), and then it won Best Paper at the conference it was accepted to.

In computer science, we are lucky to publish primarily in conferences. Journals are even worse. Best case, the turnaround is often a year or more.

Little to no change in the publication process over time.

Innovation is speeding up. As we continue to build on top of prior work, we find ourselves able to accomplish more with less time.

If it publishing isn’t already a bottleneck, it will be soon. As a sample point, I once published a paper that took one month to gather data and write up, and then 7 more months to publish.

What has been done in the last 5-10 years to speed this up? From what I can tell in computer science, nothing.

The major changes I have seen are that ACM and IEEE will put PDFs up online. And to help speed up dissemination of research, people will sometimes publish PDFs to their websites before the conference with a “To appear in…” before the conference title.

To channel Barack Obama, that is not change I can believe it.

A look at online publishing.

As a comparison, what has gone on in online publishing in the last 10-15 years?

It used to be that you had to rely on a publisher to get something out into the world. With the rise of the Internet, blogging platforms (Blogger, WordPress, etc.) were created which enabled anyone to publish to the web.

What about distribution? Well, email lists have existed for a while. And more recently, the big social networks popped up (Facebook, Twitter, etc.) which are great for connecting, but double as distribution platforms. The more likes or retweets a piece of content got, the further it spread into the web-o-sphere.

What about peer review? Peer review is basically a curation process. Social curation platforms such as Reddit and Digg have worked fairly well. If one isn’t a fan of social curation, personal curation platforms such as Pinterest can also be used.

There have been so many tools created for publishing, distributing, and curating content online. And innovation just keeps on happening.

Why can’t academics leverage any online tools?

One possibility.

If I were running a research group right now, I could imagine publishing and sharing without conferences and publications.

My students would each have a blog. They would use it to (1) publish philosophical thoughts on a research area to engage in online discussion, and (2) publish research findings that they want to share with the research community.

My research group would have a blog. Important posts from my students would get cross-posted to the group blog. In addition, I may write to the group blog.

We would all have Twitter accounts, and share blog posts via tweets. We would follow academics who we were interested in, and retweet posts that we liked. We would engage in short conversations over Twitter, and more meaningful conversations via blog comments or our own blog posts.

Peer review would be done with a curation platform. Similar to how conferences have a program committee, a committee of respected individuals could be in charge of curating a number of high quality posts every month. This could be manual, but could also be automatic. For example, if a certain number of committee members retweet a post, it gets automatically put into a collection of great posts.

The archival journals would be official curated collections of posts, and the citations to specific posts would be URLs.

This is just one way to do it. You could imagine many other ways. It wouldn’t be exactly the same as the current system, but it would serve the same purpose.


The benefits would be near-instant publication and distribution. Academics wouldn’t need to sit around and twiddle their thumbs for an accept or reject. They could immediately publish their thoughts and move on. At the same time, they could engage in near real-time communication and conversation about important research topics.

The result would have to be faster turn-around for research, which would speed up innovation. This sounds like a good for the world.

So, who will do this?

I would, except that I’m not in academia anymore. Maybe one day 🙂

Most people who are jaded with academic publishing have left academia. And that is problem for academia.

Would a current academic dare hop off the publication treadmill and then put their reputation on the line to experiment with creating a new publication treadmill? The problem is that professors and grad students are rated based upon their performance on the existing publication treadmill. It would be extraordinarily risky for a younger academic. If they cared about getting tenure (or graduating if they are grad students), they wouldn’t risk it.

It would have to be an older academic. The problem is that they have made their careers on the existing publication treadmill. Why change it?

So who will get things started?

It will take a brazen academic that respects the pure purpose of academia (to teach and discover knowledge), but does not respect the current incarnation of the academic system. But, what university would hire such a person?

(Photo credit:

P.S. This is post number #32 in a 100 day blogging challenge. See you tomorrow!

Follow me on Twitter @alexshye.

Check out my current project: Soulmix.

University and the academic/startup idea space

(Note: my academia-related posts are strongly colored by my experience studying computer engineering. Other fields will/may differ.)

I am not shy about my opinion that universities should be more active in the startup space.

Often, when I voice this opinion, I get the reply that academia and startups are just two separate things.

This is true.

But are they actually mutually exclusive? Or do they have a relationship closer to our friendly Venn diagram?


I would venture to say that they look closer to this. Most of you would probably agree.

This means that there is some overlap.

Furthermore, this overlap is should be very interesting to universities. Why? Because businesses become the big movers and shakers of the world. They innovate. They create jobs. They generate the alumni that universities are so proud of. And more importantly, they generate the money that gets donated back to universities.

Where academia and startups overlap.

It is clear that academics should be interested in the red circle above; that is, the ideas that are academically interesting. That is the entire point of academic research.

The next natural question is: how large is the overlap with the ideas that are viable as a startup?

I can really only speak for the tech space, but that overlap may be larger than people think it is. Google and VMware are good examples of technology startups that sprang out of academic ideas. Many people may think that the big consumer apps aren’t academically interesting, but is it really true? Twitter is a new type communication protocol. Facebook is concerned with translating human networks into the virtual world. Apps like Instagram and Pinterest are very interesting from a design/UI/UX standpoint. I fail to see how these couldn’t be interesting academically.

The main problem is that most academics don’t make it their problem to work at the intersection, and even if they do, they don’t carry the project out to the real world. It isn’t really their fault. The problem is that it isn’t rewarded within the university system. But, this is a slight tangent.

Let’s not forget about that blue circle.

That isn’t even the whole picture. As of now, we have only considered the red circle of ideas that are academically interesting.

What about the blue circle of ideas that are viable as a startup?

Yes, these ideas aren’t all interesting to technical departments…

…but what about the school of business?

Shouldn’t that be their bread and butter? Shouldn’t the school of business be interested in all ideas that are viable startups, getting startups off of the ground, and then graduating them to full-fledged real operating businesses?

Of course!

What are universities doing?

Both of those circles in the academia/startup idea space are of concern to universities.

Stanford has understood this for a while. They have (1) cultivated the startup ecosystem over generations of entrepreneurs and businesses, (2) geared many business, undergrad, graduate, and design students towards startups, (3) spun up StartX (a startup accelerator), and (4) recently begun to actually invest in startups from their alumni.

I hope that in my lifetime, a few other universities realize this and place significant resources in this direction. It would do the universities, their alumni, and the entire world a whole lot of good.

P.S. This is post number #19 in a 100 day blogging challenge. See you tomorrow!

Follow me on Twitter @alexshye.

Check out my current project Soulmix, your daily mix of food for the soul. Request an invite now for free access to the public beta!

The graduate school startup accelerator

My PhD stomping grounds where I should have been trying to build startups.

My PhD stomping grounds where I should have been trying to build startups.

(Note: my academia-related posts are strongly colored by my experience studying computer engineering. Other fields will/may differ.)

Being an academic-turned-entrepreneur puts me in the interesting position of continually second-guessing my prior academic career. In an effort to cover both sides of the story, I’ve already written a post on how my PhD has carried over surprisingly well into entrepreneurship, as well as a post on the two biggest differences between academia and entrepreneurship.

Aside from the similarities and differences, one question has constantly been on my mind.

Why couldn’t I have done a grad school for startups?

Look, I spent 5+ years as a funded graduate student. In those years, I met a lot of really good people, and was able to study almost anything that I wanted (granted that occasionally, I had to publish some research findings).

It was an intense period of time where I was learning rapidly and investing in my future.

I came out of those 5+ years with a PhD degree, a few good papers, and a gut instinct for interesting academic research that made me a candidate for faculty positions, industry research positions, and lots of other good tech jobs.

Pretty good right?

And let me re-iterate, I was completely funded for 5+ years! It wasn’t a lot of money, but enough to live a pretty good life.

Fast forward to the present.

What am I doing now?

I am 18 months into my new career as an entrepreneur. I am doing what I can to meet interesting people. I am learning whatever is necessary to create stuff, ship it, learn what works (as well as what doesn’t), and then iterate.

It is an intense period of time where I am learning rapidly and investing in my future.

It feels a whole lot like graduate school, with one big difference.

I am NOT funded. Instead, I am bleeding money.

But it is worth it. I know it is worth it because I am learning ridiculously fast. I can only imagine where I’ll be in a few years; that is, if I find a way to sustain monetarily.

So back to the question.

Why couldn’t I have done a grad school for startups?

It feels to me like a grad school which functioned like a startup accelerator would do a lot of good for the world.

I don’t mean an M.B.A. It is only 2 years, and the goal isn’t to build a startup during school. Plus, you have to pay for it.

I don’t mean a normal accelerator like Y Combinator, Techstars, 500 Startups, Angelpad, etc. These are three month programs to accelerate you to Demo Day, and then it is over. Yes, you are plugged into an amazing network, but it isn’t 5+ funded years to figure out how to do startups.

There has be something else. I’m pretty sure of it. I don’t know if it would be better off inside or outside of the university setting. But, it would be an interesting new direction for people to go (I may have to try one day when I have the resources).

I often imagine myself, and all of my academic buddies, with 5+ years of funded time to build startups. We were a good bunch, and I’m fairly confident we would have done some good shit. At the very least, we’d all gain valuable experience. But really — if we all had 5+ years to build and ship stuff? I bet at least one valuable business would have popped out the other end.

P.S. This is post number #18 in a 100 day blogging challenge. See you tomorrow!

Follow me on Twitter @alexshye.

Check out my current project Soulmix, your daily mix of food for the soul. Request an invite now for free access to the private beta!

The two biggest differences between academia and entrepreneurship


[Note: this is from my experiences in computer architecture and systems research. Your mileage may vary when applying this blog post to other fields.]

Earlier, I wrote about how my PhD unexpectedly provided a good foundation for entrepreneurship. It is almost a year later, and I still feel good about that blog post. Much of it remains true.

However, not everything carried over well from my PhD. As I’ve gotten further into entrepreneurship and the startup world, I’ve begun to realize two big differences that have been tough for me to adjust to.

(1) Time scales are much longer in entrepreneurship.

If you look at publication records these days, it is common for a great graduate student to publish 2+ top-tier conference papers per year. It is common for great professors to publish 5-10 or even more top-tier conference papers a year.

Think about this. Each paper requires a peer review, and top-tier conferences have fairly low acceptance rates (5-20%). Let’s assume a great student experiences a 50% accept rate. In order for a great grad student to publish 2+ top-tier papers a year, they are most likely working on 4+ papers a year, or at least one every 3 months.

A paper is a full production, requiring ideation, implementation, experimentation, and writing. Completing a full round in a few months is aggressive, but possible. I personally have done it, and know most of my friends with PhDs have.

This means that within academia, you are used to entire projects which turn around in just a few months. Once that one is done, it is on to the next! Things move quickly, and you get this awesome feeling of accomplishment each time you close the loop.

Startups and entrepreneurship work on a much, much larger time scale. You don’t get that feeling of accomplishment every few months. You may reach failure within a few months. But you will rarely reach success.

Finding product-market fit can take years. A great content marketing strategy can take years. Growing and scaling a product can take years.

A quick search easily confirms this. Pinterest was founded in 2009. Lyft was spun out of Zimride, which was founded in 2007. Snapchat was founded in 2011. It is now the end of 2013 and although these startups are fairly big, there is much room for growth.

Managing this difference requires managing your expectations. Understand that building a startup takes time. You better be in it for the long haul.

(2) Developing a new optimization function

Academia and entrepreneurship require completely different optimization functions.

In academia, you ask: what is novel and publishable?

In entrepreneurship, you ask: what do people want?

Being able to answer these questions requires a combination of knowledge and gut instinct (others may call it vision). Building the knowledge takes time. So does building a good gut instinct.

For an ex-academic, this difference is deadly. You spend years building your knowledge, and then years publishing. At the end of this, you believe you are smart! You think you know your stuff!

This is a huge trap. If you follow your academic instincts into entrepreneurship, you are most likely doomed for failure.

(Note: Stanford kids seem much better at this than most, and it must be because the culture is just different).

Managing this difference requires understanding this difference in optimization function. From my experience, it sounds easy in theory, but is very difficult in practice. Ignoring a trained gut instinct is tough. Developing a new optimization function is tough. Combine this with the first point, and realize it will just take time.

To you ex-academic entrepreneurs out there, good luck! I would love to hear how you are doing, and how you’ve managed the transition.

P.S. This today’s step as post 3/100 in a 100 day blogging challenge. See you tomorrow!

Follow me on Twitter @alexshye

Check out my current project Soulmix, your daily mix of food for the soul. Request an invite now for free early access to the private beta!

How my PhD unexpectedly carried over to entrepreneurship

When first quit my research job, one of my biggest fears was that I was throwing away my PhD degree.

I had spent many years in graduate school studying computer architecture/systems, learning the process of performing research, and publishing at top-tier venues. When I graduated, my career trajectory made sense. I would continue building a research career at Qualcomm Research, a leader in the mobile space that has tons of interesting problems to work on.

Quitting my job to be an entrepreneur felt like a huge career shift. In some sense, it definitely was. But as I get further as an entrepreneur (to be fair, I’m not that far), I’m surprised at how much of my PhD experience carries over to entrepreneurship. There are an amazing number of similaries.

Unstructured work.

The biggest difference between a PhD degree and most other degrees is that the PhD is unstructured. There aren’t specific assignments to follow. There are no guaranteed steps. There is no one telling you exactly what to build, how to build it, what numbers to collect, how to analyze them, etc. (Some advisors micromanage but if you do your PhD with someone telling you all the steps, I’d argue you are doing your PhD incorrectly.)

Entrepreneurship is similar. There are so many things to do, and no one to tell you exactly what to do. Sure, you can read up on all the entrepreneur porn you want (Hacker News is a great source), but nothing will prepare you for the journey. And even if you read a good bit of advice, it is from one person speaking about their limited set of experiences. There is no guarantee that any of their advice meaningfully maps to your own venture. You need to figure out what will work for you and your venture as you go.

It bleeds into your life.

During a PhD, there isn’t a clear separation between your research and your life. It isn’t like you have a homework assignment that you can complete, and then move on with your life. Your research project, more often than not, bleeds into all areas of your life. Your research becomes that problem that burns in the back of your head at all times. You go to sleep thinking about the problem. You wake up thinking about it. You think about it in the shower, at the gym, etc.

Entrepreneurship, at least at the beginning stage I am in, is exactly the same. It isn’t anything like a 9-to-5 job. I don’t really get a break from the work. And that is as I’d like it to be right now. There is so much to learn and so much to do, I feel I need to be immersed in it.

The whole product matters.

If there is any one thing I’ve learned during my PhD, it is this.

Doing good research is one thing. Getting published is something else.

Of course your want to do good research. You’d hope that good research results in getting published well. But that isn’t necessarily the true.

The reason is that a good paper is a good product. If the product isn’t good, chances of being published are low. You need to succinctly summarize and sell your work in the abstract. You need to get the reader excited in the introduction. You need to communicate well. The reader must understand your problem, your approach, and why you are making a contribution. You need the right graphs that prove your point. They should be easy to understand, present only the data necessary, and have good captions. You need to polish the product by repeatedly writing, proofreading, re-writing. There is a whole lot that goes into developing a paper.

This training crosses over well, although not completely, into entrepreneurship. Thinking about your product, and all the pieces of it, is very important. The whole package matters. How do you reach your users? How do you get their interest? Is your landing page good? Is your about page OK? Is your product intuitive and easy to use?

The process and the tools are different, but having worked on papers has crossed over well to beginning to think about entrepreneurship.

Refining your evaluation of ideas.

One of the huge benefits of doing a PhD is that you get the chance to define a research direction and a research project. During this process you learn how to come up with ideas, evaluate the ideas, and choose a project to work on, or an experiment to try.

As you progress in your research, you get better at this. You learn how to spot large problems. You get a sense for what will work, and what is publishable.

Learning to think about and evaluate ideas is valuable as an entrepreneur also. The biggest difference is in your evaluation function.

As a PhD, the question to answer is: what is novel and has research impact?

As an entrepreneur, the question is: what needs can I solve and what is viable as a business?


A large part of research is experimentation. You come up with a hypothesis, and run experiments to prove or disprove your hypothesis. You get good at coming up with a quick experiment for gathering data, looking at the data, and deciding if you should continue or you should change your research idea.

It is useful to think of entrepreneurship as experimentation also. Everything is an experiment. You try different things with your product. You try different ways targeting a specific niche of users. You try changing your niche. You try different methods of distribution. There is so much to figure out, and the only way to do it is to get something out there, give it a try, gather some data, and learn from it.

Related work search

During research, you get really good at keeping up with the state of the art. You learn to do background searches in an area, distill all the relevant works, come up with an idea, and then differentiate your approach from prior art.

This carries over to entrepreneurship to some degree. It is useful it have experience searching for related work, and thinking about how to extend that work, or differentiate from the prior work.

Entrepreneurship as research.

Paul Graham said it well in a recent essay on startup growth.

“Starting a startup is thus very much like deciding to be a research scientist: you’re not committing to solve any specific problem; you don’t know for sure which problems are soluble; but you’re committing to try to discover something no one knew before. A startup founder is in effect an economic research scientist. Most don’t discover anything that remarkable, but some discover relativity.”

There are many similarities between research and entrepreneurship. I’ve listed a few, but I’m sure there are a lot more. There are also differences, but that will be saved for a future blog post.

The point of this post is that initially, I was afraid that my PhD may have been a waste of time. Six months into entrepreneurship, I can definitely tell you that the whole experience was not a waste of time. There is a ton to learn, but the PhD experience has given me a great foundation to build upon.

Could I have learned the same lessons without a PhD, and in a shorter span of time? Surely it is possible. But that is a whole different question that I do not have a good answer to at the moment.

Six reasons universities should consider startup incubators

Matt Welsh, an academic turned Googler, recently wrote a blog post proposing startup incubators within academia. As a once-aspiring academic turned entrepreneur, this post struck a chord with me.  I agree with Matt for the most part, and wanted to voice my own thoughts.

I agree with Matt that academia is not efficient in transferring research ideas into the real world (at least from what I have seen in CS, EE, and CE). This isn’t necessarily a bad thing.  The main goal of academia should be education and research (not necessarily in that order). It should not be business or profit.  At the end of the day, if a university creates good students, and pushes the cutting edge in research, it is doing it’s job.

I also agree with Matt that it is worth asking if there is a role for an incubator within a university setting.  My inclination is that the answer is “yes”.  Here are a few reasons why I believe it may be a good idea.

1) An outlet for maximizing impact

Ask any academic about their goals, and you will hear a desire for impact.  One useful way to think about impact is along a time scale.  At one end of the scale is research that may have lasting impact decades from now.  At the other end of the scale is research that potentially has immediate impact.  There is academic research that falls at both ends of this scale, as well as everywhere in between.

If a research project has the potential for immediate impact, why shouldn’t an academic look to maximize their impact through productization?  It would be good for the world. It would be good exposure for the university. It would be good for the professors and students to transfer their research into the real world.

I’m not suggesting all research be geared towards incubators, but there is a category of university research that makes sense for productization.  In computer science and related fields, many research problems have the potential for immediate impact.  Stanford may be one of the few universities that knows this, and is well known for spinning out influential tech companies.  Luis Von Ahn has also been successful with selling ReCAPTCHA to Google, and is currently productizing some exciting research with Duolingo.  There are others cases, but productizing research is not common within academia, and perhaps it should be considered more often.

2) Provides an incentive for academics besides publishing

Academia is known as a place where you publish or perish.  Publications are the currency of the world.  When academics talk about a desire for impact, they usually mean a desire for publications as a proxy for impact.  In general, this isn’t that bad.  But, there are a few negatives that come from this.

First, it incentivizes people to work on what is publishable; not necessarily on what they believe will have the biggest impact.  It is easy for professors and graduate students to get caught in the system where they continually work on research that is novel, mainly for the sake of publication. How do I know?  I have done it myself for many of my papers. Once you learn the system, acquire a taste for what is novel, and learn how to write, it isn’t hard to game the system. Although they probably wouldn’t publicly admit it, I know many academics also do this.

Second, it incentivizes academics to leave good ideas too early. Academics use the term LPU, or least publishable unit, to describe the minimum amount of work necessary for publication.  If you know the LPU within your subfield, and want to maximize publications, it makes sense to find a hot idea, publish a LPU papers, and then leave for idea for low-hanging fruit elsewhere.  Again, I know because I’ve done it. I also know others that do it. This is unfortunate because I believe in order to truly evaluate if an idea, it requires going way beyond the LPU.

It would be good for academics to have another another proxy for impact besides publication.  The successful transfer of a research idea to the real world could be just this.  It would provide some incentive for academics to focus research away from work that is purely publishable.  It would also provide incentive to continue on research beyond publication.

3) Incubators can be profitable

This is straightforward.  It is also well known that money plays a big role in academia.  Universities need resources.  They want to up their endowment.  A university-run incubator that rolls out successful startups should also benefit monetarily.

The challenge is in running an effective incubator. Universities have an endowment to work with.  They also have plenty of brains and world-experts for research. What they are missing is knowledge that incubators have learned over the years. How do you support startups at their early stages? When and how do you suggest that they pivot? When do you launch?  How does do you manage your customers? This will not be easy and is not an area where universities have expertise.  However, it doesn’t mean that they can’t figure it out.  It is probably worth trying.

4) Growing the alumni network

High-caliber universities have high-caliber alumni networks. Graduating from the University of Illinois at Urbana-Champaign (UIUC) in Computer Engineering has given me something in common with a meaningful fraction of the engineering community. In the Bay Area, UIUC holds useful alumni networking events: I recently went to a meetup for investors and entrepreneurs.  Other schools may have similar, or better, networks.  For example, I’ve heard the Harvard Business School (HBS) network is a large reason for attending HBS.

Succesful incubators also have network effects.  Each class of startups build a camaraderie amongst each other. Over the years, the classes build a network of businesses that help each other out. It is well-known that the growing network at Y Combinator is becoming a very powerful resource to tap into. Universities with incubators can take advantage of these network effects within their alumni network.

5) Good timing

This is also pretty straightforward.  College is a great time for low-risk experimentation. College students usually do not have car payments, or a mortgage to pay. They can live off a small amount of money. It is also a great place to meet co-founders: they are everywhere!

6) Great practical education

A well-run incubator is essentially a startup bootcamp for academics.  It provides students with instant education about business plans, getting funding, monetizing, dealing with customers, testing, etc.  This education is valuable; especially for engineering students that know how to build, but are not exposed to the other stuff.

There is a movement towards viewing startups as experimental research.  The Lean Startup Movement, a popular startup methodology, is all about reducing waste by efficiently running experiments for validated learning on your customer-product fit.  Teaching students to manage these experiments would also be valuable.

In my opinion, this reason is the clincher.  University is all about education.  Even research and publication is about educating the world.  Good universities balance theoretical education with practical/applied education.  I bet that there is no better practical/applied education than bringing research into the real world. An incubator would be a great way, and perhaps the best way, for a university to make this happen.