Publishing in Computing Research

Over the past decade, I have had the privilege, and sometimes the misfortune, of publishing across several different computing communities. My work is pretty interdisciplinary, spanning networking, wireless, multimedia systems, and AR/VR, which means I submit papers to conferences such as SIGCOMM, MobiCom, INFOCOM, NSDI, IMC, ACM Multimedia, MMSys, SIGGRAPH, IEEE VR, and ISMAR. Although these conferences all belong under the umbrella of computer science, they often feel like entirely different academic worlds. They ask different questions, value different things, and sometimes seem to practice entirely different definitions of research.

The first thing that struck me was the enormous difference in review culture. Computer networking conferences (e.g., SIGCOMM or MobiCom) are relatively small communities with a narrow focus. Their papers tend to emphasize the big picture, such as why the problem matters, whether it is relevant to today's industry, and whether the ideas can realistically influence systems in the very near future. A paper can sometimes succeed or fail based on how compellingly the problem is articulated, even more than the technical details themselves.

Multimedia and AR/VR conferences operate very differently. Papers are expected to get to the technical contribution quickly, often assuming that reviewers already possess the necessary domain expertise. The reviews tend to focus on the ideas, algorithms, experiments, and technical soundness rather than spending pages debating whether the problem itself is important.

One thing that has increasingly frustrated me in networking conferences is how attached communities become to familiar applications and trends. This is specific to one of my areas which is video streaming and application layer networking. There are researchers who have spent decades working on video streaming problems, and understandably so. The area has produced excellent research and tremendous real-world impact creating an entire industry ecosystem of online video. But when entirely new forms of media emerge, such as volumetric video, point clouds, or dynamic 3D scenes, there is often resistance to treating them as first-class networking problems. They forget that we are enjoying the fruits of radical video streaming innovation decades ago.

This has always puzzled me. At the application layer, data is still data. The networking questions remain largely the same e.g., how should the network transport it, allocate resources to it, or adapt itself to new demands? Yet networking researchers sometimes become attached not to the fundamental problem, but to the particular applications they have grown comfortable with.

I have seen a similar shift in attitudes toward core networking research itself. Older generations of networking papers often introduced bold and radical ideas supported by analysis or simulations. Today, simulation-only papers are frequently met with a simple question -- "Where is the real testbed?" Testbeds are valuable, but they are not always necessary for fundamental research. Sometimes the contribution is the idea itself, or the identification of a new design space that others can later build upon.

This push toward immediate validation can unintentionally bias communities toward incremental work. If every paper must demonstrate a deployable prototype, then naturally the incentives favor solving small, practical problems over exploring risky new directions. The irony is that many of the systems we now consider foundational began as speculative ideas long before the technology existed to implement them. A field that only rewards what can be built today may slowly lose its ability to imagine what should exist tomorrow.

The biggest difference I have observed, however, is review quality. Networking reviewers often review up to twenty papers for each conference. Under those conditions, even excellent reviewers simply do not have the time to deeply understand every paper, particularly interdisciplinary ones. In multimedia and AR/VR conferences, reviewers usually handle far fewer submissions, which often allows for much more thoughtful and technically detailed reviews.

I have had papers rejected in both communities. But after reading reviews from multimedia and AR/VR conferences, I am usually left thinking, "I disagree with the decision, but the reviewers understood the paper." I cannot always say the same after networking conferences reviews. A rejection based on disagreement is healthy for science; a rejection based on a very high level review without technical details is much harder to accept.

Another aspect of networking research community that I find fascinating is its tendency to respond to dissatisfaction by creating new conferences. Networking researchers became unhappy with aspects of INFOCOM and created CoNEXT. More recently, frustrations with SIGCOMM have contributed to the creation of Nines conference. New conferences are not inherently bad, and experimentation with publishing models is important. My concern is that these decisions are often driven more by frustration with the status quo than by a long-term vision for how the community should evolve. New conferences frequently emerge around existing social networks of advisors, students, collaborators, and colleagues, which can unintentionally make outsiders feel that they are joining an already established club rather than helping build an open scientific community. From the perspective of a young researcher, the constant fragmentation can feel chaotic. Where should a graduate student publish? Which community should they invest their time in? Which venue will still matter ten years from now? Senior researchers often have the reputation and network to comfortably move between conferences, but younger researchers are still searching for a home community.

Perhaps the most troubling realization for me has been that many computing communities do not particularly respect each other's research traditions. This becomes particularly problematic during funding reviews. Networking research today is a collection of loosely connected communities like SIGCOMM, INFOCOM, CoNEXT, MobiCom, GlobeCom, ICNP, etc, each with its own culture, expectations, and definition of novelty. Researchers from one venue do not always appreciate the style of work produced by another, even when they are fundamentally studying the same problems. A paper that feels perfectly natural to a SIGCOMM audience may seem uninteresting to an INFOCOM reviewer, and vice versa. The same disconnect often appears in funding panels, where proposals end up being judged not on the merits of the ideas themselves, but on whether they conform to the norms of whichever subcommunity happens to be reviewing them. For a field that prides itself on systems thinking, we have built surprisingly fragmented research communities that often struggle to understand each other.

After nearly a decade of publishing across these venues, I have come to a simple conclusion that computer science is still a remarkably young discipline. Other sciences settled long ago on journals, editors, revisions, and continuity. We chose conferences because we wanted quick feedback and faster dissemination. Ironically, many of our conferences are now reinventing the very mechanisms that journals have used for decades. We have introduced rebuttals, revisions, reviewer continuity, resubmissions, and shepherding. We simply avoid calling them major revisions, editors, and journal processes. In many ways, our communities are all trying to solve the same problems while pretending to be inventing something entirely new. Perhaps that is simply the growing pain of a young field.

Academia vs. Industry

One question students often struggle with: should I go into academia or industry?

I’ve spent time in both, and the more I think about it, the less I believe this is a binary choice. The two paths are built on fundamentally different incentives.

Academia is driven by curiosity, asking why things work, pushing boundaries, and building knowledge that may take years to mature. Industry is driven by execution, taking ideas, shaping them into systems, and delivering impact at scale.

They optimize for different things: depth vs. speed, exploration vs. iteration, freedom vs. constraints shaped by real-world needs. And importantly, computing careers today are fluid, people move between these worlds more than ever.

I put together a short slide deck for a student talk that walks through these differences in a bit more detail.

👉 Open slides in a new tab

It's a War: None of Your Damn Business

I am an academic. My job is to understand and teach new knowledge. And lately I find it increasingly hard to sit down and do any of that without the weight of what's happening in the world pressing down on my chest. One of the problems that's causing this is war. Witnessing wars from aggressors like Russia, Israel, and the US over weak foreign states like Ukraine, Palestine, Iran, Venezuela. Civil wars in countries like Myanmar, Sudan, and many African countries. Killing hundreds of thousands of innocent civilians. All of this in the 21st century. And I'm supposed to be studying computers and publishing papers.

Since most of us are unfamiliar with this subject and it feels like it's beyond our control, what do we do? What do I do? What do you do?

Millions of ordinary citizens including academics such as myself and students, doctors, engineers, lawyers, and software developers are walking around with this same quiet unease.

Well, we must first accept that there will always be an aggressor with no morals. There will always be the strong and the weak, the advanced and the desperate. History has proven, with exhausting repetition, that at some point the powerful move against the vulnerable. Even in the 21st century. So whether we like it or not, we are never really outside of it. We are always on either the offense or defense side.

Many of us misinterpret war and defense. Fight vs. fighting back. War, the organized, financed, politically sanctioned action, where men in suits make decisions in AC rooms, and young people with nothing to lose die in the desert for it. Conflating war and defense is one of the oldest tricks in the book, and we all fall for it every time.

It's not just the violence. It's the concentration of power that makes the violence so easy to choose. A small number of people who never face real accountability make decisions that kill thousands, displace millions, and reshape the world. The rest of us find out on our phones. We are not consulted. We are, at best, informed after the fact.

Since most of us do not know how to react to these circumstances, the first thing we can do is refuse ignorance as a coping mechanism. This is, with no hesitation, meant for my academic community in particular. Ironically, we are one of the most embarrassingly ignorant communities on this subject. I understand the impulse. The news is genuinely unbearable, and there's a real argument for protecting mental health. But there is a difference between managing media consumption and simply looking away. Consuming high-quality information is helpful most of the time. Informed people are harder to manipulate, and right now the entire political economy of war runs on public ignorance and manufactured consent.

The second thing is to stop treating politics as someone else's profession. We have somehow arrived at a world where a software engineer will spend years mastering how to write code but won't spend two hours understanding how their government makes foreign policy decisions. A doctor will read fifty papers on a new treatment but couldn't name their own representative. The unseriousness of our generation, quite frankly, is embarrassing to witness. We have outsourced our civic lives entirely, and the people who filled that vacuum are not filling it well. I'm not saying everyone needs to run for office, though frankly, we need more scientists, engineers, and doctors in politics, desperately, but disengagement is simply stupidity. It is a gift to the people already in the room.

The third thing, and I say this as someone who works in technology, is to be deeply skeptical of the idea that our technological tools will save us. We are pouring extraordinary resources into technology, into the next platform, the next optimization. And I believe in that work. But technology does not have values. It reflects ours. And right now we are applying our most sophisticated tools to serving ads and maximizing engagement while using almost none of that capacity to strengthen democratic institutions, reduce corruption, or make power more accountable.

The least we can do is stay awake. And be more active. Be deliberate. Because the alternative is sleepwalking through a life scripted by others.

A Statement on Collective Human Responsibility

What is unfolding in Minneapolis is heartbreaking.

America has long presented itself as an advanced civilization and the leader of a free world. Today, it finds itself at a moral crossroads.

When a powerful force such as a regime, government, institution, or any authority with overwhelming power begins to treat everyday people as disposable, it leaves a fracture in the collective conscience. When ordinary neighbors, caregivers, mothers, fathers, and healers are met not with protection but with violence, it shakes something deep within all of us.

Two innocent people were killed in broad daylight by a powerful force acting under the authority of the American government. Renee Good. Alex Pretti. Two human beings with lives, families, and futures that ended abruptly and violently.

When innocent lives are taken, the only honest response is to mourn. To pause. To feel the weight of what has been lost. Anything less diminishes us. Under no circumstance should we allow governments, leaders, institutions, or policies to divide us from that truth. We cannot move forward by excusing violence. We cannot call ourselves civilized while defending cruelty.

I write this to my academic community because we carry a responsibility that extends beyond our work. We shape minds. We model values. We help define progress. And progress cannot mean becoming numb to violence or learning to rationalize cruelty. We cannot replace grief with justification, nor sorrow with silence.

We should not be silent. Silence does not mean neutrality. It means acceptance.

Together, we can choose a different path. We can acknowledge the pain, sorrow, and fear being felt right now. We can lead peacefully by encouraging empathy, insisting on accountability, and working to make our communities safer. We can treat one another with dignity and respect, remembering that before anything else, we are human.

This is not politics. It is not about parties, policies, or ideology. It is simply human decency.

Behind every loss is a life that mattered. If we cannot mourn innocent lives together, then we have already lost something far more precious than safety; we have lost our humanity.