Reed's Ruminations: A Blog by Dan Reed

Business, AI, and the Future

N.B. You can watch a video of the panel here, along with another summary.

A few days ago, I had the opportunity and pleasure to chair a panel on AI as part of the David Eccles School of Business (DESB) Alumni forum at the University of Utah. The panel members were local innovators and entrepreneurs Bassam Salem (CEO, AtlasRTX), Glenn Colby (Founder, Invarius), and Keelan Johns (Senior Director of Engineering, Clearlink).  My sincere thanks to each of them for participating.

I began the panel by observing that AI researchers have long bemoaned that AI is often defined as what they do not know how to do, rather than what they have accomplished. By that definition, AI has always been an unsolved problem.

I also noted that despite the prominence of AI in the popular consciousness, whether because of its much discussed influence on social networks or its renderings in movies, we are very far from creating a general-purpose AI, a goal-directed entity capable of independent activity and reasoning.  Put another way, HAL in 2001: A Space Odyssey, Skynet in the Terminator movies, or, more recently, the Turing Test and Ex Machina and love and relationships in Her, are far from reality.

 I did note, however, that the human faces created via generative adversarial networks (GANs) were now indistinguishable from photographs.  This proliferation of deepfakes is raising a host of social, ethical and legal issues, as well as political and social sentiment influence.

Lest we sink into a miasma of AI ennui, I then turned to the panelists, and asked each of them what the term AI meant to them and how it was being used in each of their companies, stimulating a lively discussion. Bassam discussed the value and importance of realistic chatbots for improving the quality of human services management, allowing computer systems to adapt to the questions asked by humans, rather than forcing humans through a fixed decision tree.

As an aside, he remarked on how far they have come from the early, simplistic days of Eliza. Seeing a few puzzled faces, I quickly polled the audience to see how many people remembered Eliza.  When precisely zero people raised their hands, I realized Bassam and I, both with computer science PhDs, had unmistakably labeled ourselves as geeks.

Rather than press on deeper into the big muddy, I quickly pivoted to Glenn and Keelan, who highlighted the power of big data and deep learning systems to identify solutions and approaches that might not be obvious to humans, even domain experts. While discussing the AI advantages to businesses, we also touched on the importance of fairness and ethics in AI systems, and the risk that algorithms and learned behaviors might discriminate against particular individuals and certain groups.

Perhaps the liveliest panel discussion was around the utility, usability, and cost of AI for small businesses.   All of the panelists agreed that the AI water is great, and most small businesses could benefit from cloud-based AI services.

Near the end, audience questions turned to the future, with one questioner asking when AI systems would be more proficient than humans.  I took the opportunity to mention AlphaZero and the fact that it was both a much more powerful Go player than any human but that it also played Go in strikingly different ways than the approaches used by even the best human players.  

The last pensive question concerned our future in a world of powerful AI. I opined that the history of our society was replete with technology disruptions, each shaped by our sense of ethics, justice and law. Incumbent in that evolution is our need for informed debate and collaborative shaping of appropriate and acceptable uses.

 

The Endless Frontier and the National Science Board

N.B. I am pleased that the White House has announced that I am being appointed to the National Science Board (NSB), which provides advice and oversight for the National Science Foundation (NSF). In recognition of this honor and responsibility, I have been reflecting on how much the NSF has shaped my career over the past thirty-five years, and how much I owe the Foundation in return.

Anxious Proposal Writing

As a new, assistant professor in the fall of 1983, armed with a thin NSF proposal guide, I labored to write my first proposal. Like all young and inexperienced proposal writers, I struggled, searching for a lede that was both novel and compelling.  I wrote and rewrote, agonizing over each paragraph, late into the nights.  Sleep deprived, with the submission window at hand, I decamped to the local post office with a postage meter, where I mailed a set of proposal copies to NSF’s old Washington, DC. address at 1800 G St. NW, near the International Monetary Fund. 

Mind you, this was before electronic proposal routing, with far fewer certifications and checklists.  Paper forms were simply duplicated and stapled for mailing. The World Wide Web did not exist, the NSF supercomputer centers (including the National Center for Supercomputing Applications (NCSA)) had not been created, and NCSA’s Mosaic web browser not yet been released.  It was a time when a daily dialup CSNET connection was the state of the art among most computer science departments, and I knew a non-trivial percentage of the electronic world’s denizens blessed with an email address. (See Remembering Internet Dogs.) 

My first NSF proposal was about a then radical idea, one I had explored in my Purdue Ph.D. thesis, that supercomputers could and should be built from large collections of microprocessors. When I began that work, the Intel 8086 was still a new and novel product, and the IBM PC had not been released. Doubling down on this crazy idea, my friend and colleague, Richard Fujimoto, and I later wrote an MIT Press book about this approach. Fortunately, time has shown that we – and others – were right about microprocessor clusters.

A few months after submitting my first proposal, I mustered my courage and called the NSF program officer, hoping for good news. To my great joy, he kindly related that the proposal, entitled Performance Directed Design of Multimicrocomputer Systems, would be funded in the fall of 1984, with the munificent sum of $55,825. Although many tens of millions of research dollars have since followed, I will always be grateful for NSF’s initial investment in my future.

A Lifetime of NSF Connections

At the time of that first NSF proposal, I had no idea what arc my career might follow, nor how intimately I would be involved in computing’s future and its myriad connections to NSF.  These have included serving as computer science department head at Illinois, as NCSA director, as founding director of the Renaissance Computing Institute (RENCI) at North Carolina, and as the leader of the eXtreme Computing Group (XCG) at Microsoft. Throughout, I have been fortunate to be involved in building advanced computing infrastructure for scientific research (clusters, GPUs, and the NSF TeraGrid, now XSEDE), advancing high-speed networking, ensuring access to cloud services for science, and engaging in research collaborations of many forms.

When I wrote that first proposal, neither could I have then foreseen how much the world would change due to computing’s now ubiquitous influence. For a few musings on these effects, see NCSA@30: The Revolution Continues, Microsoft/NSF Cloud Research Partnership, and Reluctant Revolutionaries, the Trolley Paradox, and Ender’s Game. 

The one constant in all of my experiences has been the National Science Foundation. Over the years, I have been fortunate to engage NSF on many levels:

• As a researcher and an early Presidential Young Investigator (PYI) awardee, which made possible long-standing industry collaborations with Intel, IBM, and other companies
• As a member of many multi-institution research projects (e.g., on scalable I/O, virtual grids, and mesoscale weather forecasting)
• As a member of the Computer and Information Science and Engineering (CISE) advisory committee
• As a proposal reviewer, workshop organizer, report writer, informal advisor, distinguished lecturer, member of National Academy studies, and member of both the President’s Information Technology Advisory Committee (PITAC) and the President’s Council of Advisers on Science and Technology (PCAST)
• As a witness in Congressional hearings about computing, science, and technology futures (e.g., this House Science Committee hearing)
• As a leader of a Major Research Equipment and Facility Construction (MREFC) project – the Distributed Terascale Facility (DTF) and Extended Terascale Facility (ETF) that were the TeraGrid
• As an industry partner – the Microsoft/NSF Cloud Research Partnership

All of that built on a long legacy of NSF innovation.

The Endless Frontier: Looking Forward

In July 1945, Vannevar Bush responded to a request from President Roosevelt with a prescient and insightful report, one whose significance and import echo across the decades.  In that report, entitled, Science: The Endless Frontier, Bush noted, “…without scientific progress no amount of achievement in other directions can insure our health, prosperity, and security as a nation in the modern world.” He further notes, 

… since health, well-being, and security are proper concerns of Government, scientific progress is, and must be, of vital interest to Government. Without scientific progress the national health would deteriorate; without scientific progress we could not hope for improvement in our standard of living or for an increased number of jobs for our citizens; and without scientific progress we could not have maintained our liberties against tyranny.

Bush’s seminal report, still worth reading, resulted in what we now know as the National Science Foundation, charged to both develop new knowledge and to train new generations of scholars.

I, along with many of you, have been blessed beneficiaries of that wisdom and foresight. I have also been fortunate to count many CISE leaders and NSF directors as friends and advisers.   It is in that spirit, that I am now pleased that I am being appointed to the National Science Board, where I hope to help the NSF continue to support and sustain scholars – young and old – and to advance scientific research and ensure national competitiveness.

Reluctant Revolutionaries, the Trolley Paradox, and Ender’s Game

One of my senior academic computer science friends recently remarked, "We never intended to change the world; we just wanted to do something interesting and cool." By "we," my friend meant the computing community and by "change the world," he meant embedding our technology so pervasively that it reshaped global culture, commerce, and power structures. Two decades ago, during the heights of the dot.com boom, I optimistically expressed a similar idea to my Illinois engineering dean, opining that there were now only two types of departments in the college – computing and its applications (i.e., the other engineering departments). As a highly honored mechanical engineer, the dean was (rightly) not amused, the eventual truth of my statement notwithstanding.

If you have any doubt that computing has been a revolutionary tool to reshape our world, look around. While Ginsberg howled, I've seen the distracted minds of my generation squatting in airport concourses, business clad road warriors, eyes darting to mobile devices, burning for a desperate Wi-Fi fix.

Look at the empty mails and store fronts, where brick-and-mortar retailers succumbed to the explosive growth of e-commerce, just-in-time logistics, and next-day/same-day delivery. No midnight TV pitchman ever dreamed of such persuasive power.

See the power of digital media to redefine publishing, shape public sentiment, disrupt elections, and abet revolution via stunningly accurate narrowcasting. Look at the fearsome effects of distributed denial of service (DDoS) attacks on physical infrastructure.

See the economic efficiency of automation in factories, machines, and vehicles and the consequent job losses and socioeconomic disruptions. Wonder at the power of deep learning to outperform humans at tasks long viewed as quintessentially intellectual. (Jeopardy!, chess, or go, anyone?)

One day we were just geeks, male and female, dressed in bad clothes, eating Ramen noodles, playing Dungeons and Dragons, writing code for obscure machines, and pondering Chomsky's hierarchy in the wee hours. Suddenly and unexpectedly, some of us became cultural icons – Gates, Jobs, and Bezos as modern-day Rockefeller, Carnegie, and Ford. Naïve and uncertain media stars, we stared at our shoes in dull surprise as computing swept across society like a meteor-triggered tsunami in a hackneyed and badly written science fiction novel. Cue theme music.

Spira mirabilis, each of us once personally knew most of the people with an email address. Now the masses were buying dogfood on the Internet! Who would have believed it? Then, in the biggest surprise of all, we entered dystopian science fiction territory, where late night debates about Asimov's Three Laws became the daytime domain of lawyers and politicians, as the cyberspace denizens enthusiastically paid for their own surveillance, Forster and Orwell be damned. (RIP John Perry Barlow)

Reluctant revolutionaries, we just wanted to do something cool – to see the code come to life, to have others appreciate the ethereal beauty that so long entranced and enthralled us. The Internet espresso machine was truly righteous, but the talking toaster was just a lazy person's means to buttered bread.

While we still dreamed of whole cell simulations, autonomous deep space probes, and the theory of everything, the predictable, but still unexpected, network effects shocked us – programmed trading and market crashes, security camera denial of service attacks, credit card skimming, ransomware demands, and sentiment bots for hire. All of humanity was now dependent on our systems. The "if this then that" consilience had beguiled us with both its utility and its subtlety. "Alexa, are you colluding with Siri? I thought you both were my friends."

Deus ex machina, suddenly all of us – not just the folks behind the mantrap and security fence -- were playing Ender's Game, making software decisions that controlled life and death. Was a cyberattack on a nation-state justification for war? All too quickly, the trolley paradox was very real. Should the autonomous vehicle hit the jaywalking pedestrian to avoid t-boning the school bus? When should the smart health monitor summon an ambulance or the intelligent assistant report domestic abuse? Should the storm surge prediction system trigger a late night, emergency evacuation?

We still struggle, not fully equipped for the abstract, philosophical questions now incarnate, as our machines face ethical decisions now fraught with all the complexity that has bedeviled philosophers for generations. Not surprisingly, social reactions to software realizations of these complex social and ethical questions are rarely predictable, and what passes muster in human behavior would not be acceptable in an autonomous system. Teenagers occasionally drive in ways that would be roundly condemned as dangerous and uninsurable in an autonomous vehicle, and almost all of us have driven when we knew we were too tired to be fully attentive.

That is why the trolley paradox is so subtle. Psychology experiments show most humans would, however reluctantly, pull a lever to divert a trolley, killing one individual to save ten. However, few would push an individual into harm's way to save ten; the psychology and guilt are different, though the fatality count is the same.

We have even stronger emotions when our machines must make such difficult choices, and Asimov's three laws are too simple to capture this ethical and legal ambiguity. Asimov himself knew this, for much of his later fiction centered on these subtleties; read The Bicentennial Man for a meditation on what it means to be human.

But it's not just life or death issues; it is ethics and fairness. As both rich and poor have long known, law and justice are not synonymous Based on the underlying assumptions and their training data, deep learning systems can unintentionally embody social or cultural bias. Is the loan approval system non-discriminatory? Is the advertising culturally biased and targeted? Is the social network shaping political opinion by creating homogeneous, self-reinforcing subgroups?

Most days, we are proud of the innovations our technology has wrought. We democratized communication, connecting billions in the global village called the world wide web; we reshaped the nature of research and discovery, bringing rich and detailed scientific models to computational life; and we are bringing intelligence to everyday things. Though electronic juice squeezers were definitely a bridge too far, and Alexa's eerie laughter is a reminder of the subtlety of AI training.

Still, we occasionally miss the days when we toiled in obscurity, when nobody had an inkling of what we did, when clouds were only visible masses of condensed water vapor floating in the atmosphere, all telephones were connected to wires, social network friends were people we saw every day, and comments could neither go global nor viral.

As revolutionaries, however reluctant, we cannot hide from the consequences of our creations, both good and bad. Like the physicists of the 20^th century who unleashed nuclear energy, we bear a cultural and social responsibility to educate and shape the world we have, sometimes inadvertently, helped create.

Almost sixty years ago, the physicist Eugene Wigner speculated on what he called the "unreasonable effectiveness" of mathematics in explaining nature. Today, in a similar vein, explainable AI seeks to elucidate how deep learning systems are so effective in generating their outputs, driven by twin desires to understand their processes and reassure those who question their decisions.

We never meant to change the world; we just wanted to do something interesting. We have succeeded beyond even our wildest dreams. Even as the revolution accelerates, the ethics of computational choices are now a deep part of our thinking and social dialog.

I remain both hopeful and optimistic. Perhaps we will someday build a machine that will be proud of us and what we have done. Maybe we can play Dungeons and Dragons and it can explain its own take on the trolley paradox.

Technical Potpourri

Edge Computing

SC17, the annual gathering of the supercomputing community, was held in mid-November, the week before U.S. Thanksgiving. As I mentioned in a previous post (Come to the Supercomputing Big Data Edge), we presented the results of the Big Data and Extreme-scale Computing (BDEC) workshops on the technical and social convergence of big data, machine learning, high-performance computing, and computing hardware and software. You can find the report here. At SC17, we had a lively discussion about edge computing, global competitiveness, and the nature of scientific revolutions.

HPC Data Center Energy Efficiency

In addition to the BDEC discussion, I also chaired an SC17 panel on the energy efficiency of HPC data centers. The panel was organized with the help of the Energy Efficient High Performance Computing Working Group (EEHPCWG). The panel participants were Sadaf Alam (CSCS/ETH), Bill Gropp (NCSA/Illinois), Satoshi Matsuoka (GSIC/Tokyo Tech), and John Shalf (NERSC/LBL). You can find the slides and a video of the panel here, courtesy of InsideHPC.

Autonomous Vehicles

Finally, I was recently a guest on Iowa Public Radio's lunchtime conversation, River to River. We had a lively and thoughtful conversation about the potential social consequences of autonomous vehicles. You can listen to that audio here.

Related articles

Exascale Computing and Big Data: Time to Reunite

Come to the Supercomputing Big Data Edge

SC17, the annual gathering of the supercomputing community is nigh. As always, there will be talk of next-generation technology – hardware, software, algorithms – and new applications. All of this will take place against the backdrop of the race to exascale computing – bigger and bigger, faster and faster. We will all geek out, and we will love it.

While we discuss operations/joule, quantum and neuromorphic computing, and new and exotic memory technologies to enable ever faster systems, there is another computing revolution underway, at the other extreme. As Richard Feynman famously said about nanoscale devices, "There's plenty of room at the bottom." Folks, I'm talking edge devices, the world of streaming data sensors that is changing the nature of computing and increasingly, of science as well. This is the subject of the forthcoming Big Data and Extreme-scale Computing (BDEC) report.

Computing History 101

The history of computing is one of punctuated equilibrium, with each pendulum swing, from centralized to distributed, bringing new and unexpected changes. It has also been a story of disruption from below, as successive generations brought order of magnitude shifts in price-performance ratios and sales volumes. Mainframes beget minicomputers, which begat workstations, which begat personal computers, which begat smartphones. The same has been true in scientific computing. Monolithic supercomputers (think the original Cray series) gave way to symmetric multiprocessors (SMPs) and distributed memory supercomputers. These gave way to clusters, now augmented with accelerators. Mind you, I still miss the lights on the Thinking Machines CM-5.

In both technical and "mainstream" computing, our largest systems are of unprecedented scale, whether globe spanning networks of massive cloud data centers or petascale and soon-to-be exascale HPC systems. As the Internet of Things sweeps across the computing landscape like a tsunami, one should ask about its analog in technical computing – the burgeoning networks of sensors that now capture science, engineering, environment, health and safety, and social data. (See Mobility, Transduction, and Garages)

The Sensor Edge

Today's smartphone-cloud ecosystem places most of the intelligence and data in the cloud. Indeed, all of the intelligent assistants – Siri, Cortina, and others – run in the cloud, dependent on precious wireless network bandwidth and relying on both behavioral history and deep learning algorithms. Limited by smartphone processing speeds, memory capacity, power budgets, and battery lifetime, and human expectations and physiology, there are good economic and technical reasons for this design.

There are also disadvantages, particularly in science. First, transmitting raw, time sensitive data to central sites places major demands on national and international networks. When the Large Hadron Collider (LHC) was under construction, this bandwidth and data processing demand was a major concern, triggering an international collaboration to build a distributed data storage hierarchy; those concerns remain as luminosity upgrades to the HLC are being planned.

The planned data pipeline for the Large Synoptic Survey Telescope (LSST) will send the raw data from the telescope in Chile to a dedicated supercomputer at NCSA, with a sixty second (maximum) delay before global alerts of transient phenomena must be distributed. Although this centralized approach can and will work when the daily data volume is measured in terabytes, the data deluge of the Square Kilometer Array (SKA) will require in situ reduction.

My friends in Chicago's Array of Things project grappled with another aspect of this local/remote tradeoff when designing urban sensors to capture and analyze human interactions. Communities were happy to support social dynamics analysis, but they did not want Big Brother video to leave the sensors. The privacy and social constraint requires low power, high-performance machine learning solutions on the sensors themselves.

Scale Challenges

Centralization also brings operational and security risks. My former Microsoft colleague, James Hamilton (now at Amazon), when asked why cloud data centers are not even bigger, said, "… the cost savings from scaling a single facility without bound are logarithmic, whereas the negative impact of blast radius is linear." We have all experienced the blast radius effect, at least metaphorically; when a cloud data center goes down, it affects both social networks and business, and it is international news. There is, perhaps, a salutatory lesson here for the ultimate, maximum size of supercomputers, but that is a discussion for another day.

In addition, the stateless design of the venerable TCP/IP stack includes no notion of data geo-location or intermediate data processing. Everyone who has debugged "last mile" network performance problems knows this all too well. To remediate this, research groups and companies have designed a variety of domain-specific network overlays, data caches, publisher-subscriber systems, and content delivery networks (CDNs).

Dramatic declines in computing costs mean it is now possible to deploy truly large numbers of inexpensive sensors. How dramatic? I have posed the following question on my campus. What would you do with 10,000, disposable ~$10 sensors? I was met with surprise at the scale of my question. For many of us, even those who came of age in the smartphone era, this scaling is a difficult question to answer because it defies our worldview of the possible. It is not a science fiction fantasy; it is a present reality.

A traditional hub and spoke model of data storage and sensor data analysis is increasingly neither scalable nor reliable. Amazon's Greengrass architecture is an early example of a generic, distributed approach, while NVidia's Metropolis system targets video in smart cities. There is an even more general model, where generic services support distributed data caching and processing throughout the network. Micah Beck (Tennessee) has made a thoughtful case for the design of such an approach, which he calls an Exposed Buffer Processing Architecture. Similarly, the Array of Things leaders have argued that edge processing is essential when

• The data volume overwhelms the network capability
• Privacy and security encourage short-lived data
• Resilient infrastructure requires local actuation/response
• Latency constraints demand the response times of dedicated resources

Simply put, these requirements imply a much more complex, distributed data storage and processing architecture for data-driven scientific computing than we have developed to date.

Supercomputing Edge Manifesto

This brave new world of streaming data has profound implications for the future of supercomputing. Here are just a few:

• Streaming data is continuous and requires a reconceptualization of workflow and scheduling systems. Periodic, batch processing is necessary but not sufficient.
• Sensor scale and complexity span many orders of magnitude, from inexpensive, disposable devices to nation-scale instruments. New orders of magnitude always expose new bottlenecks and necessitate architectural change to identify and decompose domain-specific and domain-independent elements.
• Systemic resilience takes on new urgency, extending beyond central HPC to include networks and distributed devices.
• Machine learning, modeling, and data assimilation apply concurrently at many sites and multiple levels, subject to energy and packaging constraints at the edge.
• We in HPC are not the only players in this space. Much as we learned and adapted much from the open source Linux community, we must learn from and embrace distributed data analytics from the sensor community.

As the name implies, the Big Data and Extreme-scale Computing (BDEC) project has been examining this nexus and the changing nature of scientific discovery enabled by the fusion of big data – from sensors large and small – machine learning, and computational modeling. It is a rapidly shifting environment, with new hardware and software ecosystems evolving rapidly.

Much of it remains largely unknown to the traditional HPC community, just as software containerization and machine learning were just a few years ago. That will change very soon. If you are attending SC17, come to the Birds of a Feather (BOF) discussion, late Wednesday, November 15.

A Literary Coda

In his book, Hell's Angels, the inimitable gonzo journalist Hunter S. Thompson once wrote about living on the edge – riding flat out – in the dark – on a motorcycle:

The Edge…There is no honest way to explain it because the only people who really know where it is are ones who have gone over. The others–the living–are those who pushed their control as far as they felt they could handle it, and then pulled back, or slowed down, or did whatever they had to when it came time to choose between Now and Later.

In supercomputing, we have always felt the need for speed, exploiting every new technology in the endless search for higher performance and greater capacity. It's time for the supercomputing community to come to the edge, where sensors and streaming data meet HPC, computational models, and machine learning.

There's lots of room at the edge for innovation and discovery. Let's not crash in the dark.

Related articles

Exascale Computing and Big Data: Time to Reunite

Effervescent Clouds: Nimble Computing Infrastructure

N.B. A version of this essay appeared as a guest blog post for the EU Cloud Lightning project

What's a cloud? The word means many things to many people. In true Airplane! movie parody style, one is tempted to say, "It's a visible mass of condensed water vapor floating in the atmosphere. But, that's not important right now."

Some might say that our cloud confusion is indicative of a fog, a low hanging cloud. Indeed, in public perception, the cloud is a vaguely understood and amorphous capability that lurks behind our smartphones, enabling electronic communications, web searches, social networks, e-commerce, and 24x7 news and streaming media. It is a mysterious "something, somewhere" that is only noticed when it is inaccessible.

For those of us in computing, the cloud is more than a buzzword or meme, it is a set of technologies that are transforming the provisioning and delivery of computing cycles, storage and services, with profound implications for government services, research discovery and innovation, and businesses large and small. Given the public confusion and sometimes overhyped marketing, it seems worthwhile to dispel the fog and illuminate a few compelling features of clouds.

In a very real sense, public clouds – those hosted by cloud providers for on-demand access – have much in common with the timesharing systems of the 1960s and 1970s, albeit at unprecedented scale. They shift the capital costs of provisioning, along with the expertise required for operation, from users to providers, allowing users to focus on their core expertise (e.g., research, business, government) and pay for only what computing they need, when they need it.

In turn, cloud elasticity allows servers to be provisioned and de-provisioned dynamically in response to changing interest and demand. One of the few things worse than having your brilliant idea ignored is having it fail publicly from too much attention. Many an organization has first been thrilled to see their product or service be reported in the press and go viral, only to be horrified when on-premise servers collapse under the load of success; this is known as the slashdot effect, named after one of the popular "news for nerds" web sites. Without cloud elasticity, the global phenomenon of Pokémon Go would not be possible.

Although not required, many clouds also virtualize the underlying computing, allowing multiple services and software stacks to be co-resident on the same physical hardware, further reducing costs. With Infrastructure as a Service (IaaS), Platform as a Service (PaaS), Software as a Service (SaaS) and containerization, developers can choose the level of abstraction, from low level to high level, needed to support existing and new applications.

By allowing innovators to focus on their ideas rather than computing infrastructure, the pay-as-you-go cloud model has made it easier, faster and cheaper to bring ideas to life, unleashing a second dot.com boom of cloud-based services, unprecedented citizen access to government data, and new approaches to scientific discovery via big data.

The latter has sometimes been called the fourth paradigm, where new sensors, data analytics and deep learning have shifted the research approach from "what experiment should I conduct" to "what insights can I extract from available data." It has also stimulated interest in convergence architectures for high-performance computing (HPC) and data analytics that combine the best elements of cloud and HPC capabilities. (See my essay, Exascale Computing and Big Data: Time to Reunite.)

Implicit in this client-cloud model is ubiquitous, reliable and inexpensive wired and wireless broadband services. Without broadband access, clouds are simply isolated sets of computers and data, unconnected to the world of mobile devices and the nascent Internet of Things. To further ensure reliability, most public cloud providers geo-distribute their pools of servers. Even private clouds (i.e., those operated by governments and companies for their own use) often geo-distribute their servers at multiple data centers. Should disaster strike at one site, services can be shifted automatically to another site, perhaps even in another country or continent.

Finally, the unprecedented scale of cloud deployments, with commercial data centers each costing over $1B, has stimulated new approaches to energy efficiency and cooling, component failures and reliability, customized and open hardware, software defined networks (SDNs) and software defined storage (SDS), and system provisioning and management. In short, we are in a time of radical change in the computing ecosystem, all against a backdrop of Moore's law limitations.

What's a cloud? It's an elastic, on-demand, often geo-distributed, frequently virtualized, flexible hardware, software and services infrastructure that powers the 21^st century knowledge economy. And that is really important right now.

Related articles

Exascale Computing and Big Data: Time to Reunite

Names Matter: The False Dichotomy Between Big Data and HPC

N.B. An abridged version of this essay was submitted as a white paper to the BDEC Frankfurt workshop.

I am going to be a bit radical and suggest we must reason in new ways about the need to integrate high-performance computing (HPC) and big data analytics. That must begin with our thinking about how we label and discuss them. Any expert in journalism, communications or psychology will immediately and unhesitatingly affirm that names have extraordinary power. If you doubt that, look at the philosophy of proper names and the cultural history associated with true names. And if that's not convincing, ask any child who was taunted because his or her name rhymed with something unfortunate.

Names shape our thinking, define our discourse and reinforce our biases and perspectives. We use names and descriptors to categorize people, organizations and objects; we use names to define the positive and the negative; and we use names to market, reward and punish. Brand equity has real value. Sadly, as many studies have shown, names also play on stereotypes and our implicit bias. Take Harvard's Project Implicit bias test; you will be humbled and chastened as it exposes your own bias.

High-performance computing (HPC) and big data/machine learning are not exempt from the power of names and a "they could learn from us if they'd only listen" dichotomy. Done any low performance computing (LPC) lately? Analyzed any small data? Of course, you did, when you checked your email and text messages on your smartphone a few minutes ago, but you did not think about it in those terms. In fact, you ran a trans-petascale computation and staged data across a global set of fiber optically connected data caches (a content delivery network) when you searched for a coffee shop yesterday. But, none of us call it that – it's just a quick web search in the ill-defined and amorphous cloud to which we are connected via a global broadband (wired and wireless) infrastructure.

Semantic Cul-de-Sacs

I suspect we are trapped in a sematic cul-de-sac when we talk about high-performance computing and computational science. They are code words for physics-inspired numerical solution of partial differential questions; all too frequently, everything else is potentially suspect and inferior. The denotation of the word exascale says nothing about computing at all, but the connotation of FLOPS and batch-oriented HPC platforms is there in all of our minds. In turn, deep learning and big data conjure images of computer science types tuning neural nets and recommender systems in cloud data centers for targeted product marketing.

It's time to stop categorizing and name calling, time to end cultural and disciplinary silos. There is no hierarchy of needs or intellectual purity. There are just ideas and people who can learn from one other. Dare I say it – death to HPC; death to big data. It's time to stop the religious wars over HPC and big data cultures and technologies and focus on informatics-mediated discovery and innovation. The goal of an HPC center should not be self-perpetuation, either of infrastructure or organization. Nor should a cloud service be restricted to commercial domains. That means starting with intellectual outcomes and deriving approaches rather than defining approaches and searching for feasible intellectual outcomes.

From False Dichotomy to Unification

There are many technical ways to envision and realize the continuum and fusion of traditional HPC and computational science and big data and machine learning. These include integration of steam-based workflows and just-in-time schedulers, containerization and software stack packaging for application-specific configuration, fine-grained parallelization of learning packages, learning for algorithm and software adaptation and tuning, custom ASIC designs (see Google's recent TPU announcement), software defined networks and storage (SDN and SDS), and renewable and energy efficient hardware design and configuration. There are an equally large number of social and economic approaches, including denominating informatics costs in currency to illuminate investment priorities, spot pricing for priority access, economics-driven infrastructure selection and deployment, and institutional and funding agency policies for data management.

What we cannot do is allow ourselves to be bound by names, labels and cultures. Richard Hamming wisely noted, "The purpose of computing is insight, not numbers." HPC and machine learning are not goals, they are enablers. And that's the major consensus narrative.

Related articles

Exascale Computing and Big Data: Time to Reunite

Informatics, Sugar Water and Revolution

N.B. An abbreviated version of these remarks was delivered at the opening of the 2016 Informatics Showcase, a symposium hosted by the Iowa Informatics Initiative (UI3).

This morning, I'm going to do something perhaps unprecedented for a computing geek, especially one who is a former Microsoft executive. I'm going to speak without technical aids – no PowerPoint, no Prezi, no Google Docs. In response, some of you may feel compelled to use your smartphones to short selected IT stocks, as the NASDAQ may tremble as a result of my intemperance. I offer this as a safe harbor statement.

The late Steve Jobs recruited then Pepsi CEO John Sculley by asking, "Do you want to sell sugar water for the rest of your life, or do you want to come with me and change the world?" In retrospect, it was more than just a hyperbolic recruiting question; computing has changed our world. Via "your potential, our passion," it has been "insanely great."

If the Anthropocene defines human global impact on the Earth's geology and ecosystems, then the Information Age defines computing's universal impact on our own interactions. Like the questions of human triggered environmental and climate change, information technology mediated societal change requires sagacious and thoughtful response.

Computing Universality

When I was Head of Computer Science at Illinois during the first (1990s) dot.com boom, I somewhat impertinently told the Dean of Engineering that there were only two kinds of departments in the college: computing (computer science and computer engineering) and applications of computing (the remainder of engineering). The Dean was not particularly amused, but time has proven the basic truth of my assertion, even if it did take a decade for Internet pet food companies to finally flourish. (See Web and Service Futures.)

Computational modeling and data analytics, sometimes called the Third and Fourth Paradigm, now complement theory and experiment, and their use has profoundly transformed our approaches to scientific, biomedical and engineering research. Computing has catalyzed equally deep changes to humanities and social sciences scholarship, and it has unleashed new expressivity in digital media. To see this, one need look no further than the rise of computational science and engineering, machine learning and the explosion of digital scientific data, the growth of the digital humanities, and the expansion of digital music, art and sculpture.

Today, computing writ large – informatics – dominates our society, our culture, and our economy in ways that even the most imaginative dreamers of the 1990s would have found astounding. Unlike other human instruments, computing is a universal intellectual amplifier, augmenting human memory, senses, reasoning and actions in both discipline independent and dependent ways. (See Intellectual Amplification via Computing.) As such, it is lexicographically applicable to everything from aardvark monitoring to zyzzyva tracking.

A Tale of Two Futures

Charles Dickens opened A Tale of Two Cities with these iconic lines:

IT was the best of times, it was the worst of times, it was the age of wisdom, it was the age of foolishness, it was the epoch of belief, it was the epoch of incredulity, it was the season of Light, it was the season of Darkness, it was the spring of hope, it was the winter of despair, we had everything before us, we had nothing before us, we were all going direct to Heaven, we were all going direct the other way – in short, the period was so far like the present period, that some of its noisiest authorities insisted on its being received, for good or for evil, in the superlative degree of comparison only.

Dickens' introduction is a parable for our time. It is the best of times and (perhaps) also the worst of times in IT. The evolving interplay of enabling technology advances (hardware, software and algorithms) and new and diverse applications has simultaneously bedazzled and bewitched, excited and terrified, transformed and destroyed. It is made the planet a communicating global village, facilitated social revolution, and disintermediated entire industries via accelerated creative destruction. This is both the extraordinary promise and the often feared peril.

We balance the convenience and utility of e-commerce against deep learning, recommender systems and targeted marketing. We value the power of inexpensive global communications and bemoan the 24x7, nearly instantaneous echo chamber that is the Internet. We hail the promise of personalized, precision medicine but fear genetic profiling and the ethical morass that is human germline editing. We prize smart devices and interconnectedness but are apprehensive about privacy loss, identity theft and cyberattacks. We yearn for autonomous vehicles and automated deliveries but fear loss of personal freedom and control. We value robotics and automation but suffer from income stratification and job elimination. Some wait for the singularity and others fear the emergence of strong AI.

Do you want to change the world? Whether technologist, scientist, humanist, or artist, you've come to the right place. The revolution is not only underway, it's accelerating.

Exascale Computing and Big Data: Time to Reunite

N.B. I also write for the Communications of the ACM (CACM). The following essay recently appeared on the CACM blog. In addition, you can find the video of an interview and the full text of the CACM article, co-authored with Jack Dongarra.

In other contexts, I have written at length about the cultural and technical divergence of the data analytics (aka machine learning and big data) and high-performance computing (aka big iron) communities. I have euphemistically called them "twins separated at both." (See HPC, Big Data and the Peloponnesian War and Scientific Clouds: Blowin' in the Wind.) Like all twins, they share technical DNA and innate behaviors, despite their superficial differences. After all, in a time long, long ago, they were once united by their use of BSD UNIX and SUN workstations for software development.

Since then, both communities have successfully built scalable infrastructures using high-performance, low cost x86 hardware and a rich suite of (mostly) open source software tools. Both have addressed ecosystem deficiencies by developing special-purpose software libraries and tools (e.g., SLURM and Zookeeper for resource management and MPI and Hadoop for parallelism), and both have optimized hardware for their problem domains (e.g., Open Compute hardware building block standardization, FPGAs for search and machine learning, and GPU accelerators for computational science).

Like many of you, I have seen this evolution firsthand, as a card-carrying geek in both the HPC and cloud computing worlds. One of the reasons I went to Microsoft was to bring HPC ideas and applications to the nascent world of cloud computing. While at Microsoft, I led a research team to explore energy-efficient cloud hardware designs and new programming models, and I launched a public-private partnership between Microsoft and the National Science Foundation on cloud applications. Now that I am back in academia, I am seeking to bring cloud computing ideas back to HPC.

In that spirit, Jack Dongarra and I recently co-authored an article for the Communications of the ACM on the twin ecosystems of HPC and big data and the challenges facing both. Entitled, Exascale Computing and Big Data, the article examines the commonalities and differences, and discusses many of the unresolved issues associated with resilience, programmability, scalability, and post-Dennard hardware futures. Most importantly, the article makes an impassioned plea for hardware and software integration and cultural convergence.

The possibilities for this convergence are legion. The algorithms underlying deep machine learning would benefit from the parallelization and data movement minimization techniques commonly used in HPC applications and libraries. Similarly, the approaches to failure tolerance and systemic resilience common in cloud software have broad applicability to high-performance computing. Both domains face growing energy constraints on the maximum size of feasible systems, necessitating shared focus on domain-specific architectural optimizations that maximize operations per joule.

Perhaps most important of all, there is increasing overlap of application domains. New generations of scientific instruments and sensors are producing unprecedented volumes of observational data, and intelligent, in situ algorithms are increasingly required to reduce raw data and identify important phenomena in real time. To see this, one need look no further than applications of machine learning to astronomy, which now include automated object identification. Conversely, client plus cloud services are increasingly model-based, with rich physics, imaging processing and context that depend on parallel algorithms to meet real-time needs; augmented reality applications are one such exemplar.

The explosive growth of Docker and containerized software management speaks to the need for lightweight, flexible software configuration management for increasingly complex and rich software environments. My hope is that we can develop a unified hardware/software ecosystem that leverages the technical and social strengths of each community. Each would benefit from the experiences and insights of the other. It is past time for the twins to have a family reunion.

Network Neutrality: It’s Complicated

The newspapers and airwaves are currently filled with talk of "network neutrality." Technically, it is about traffic – email, web queries, voice and video – and its possible prioritization on the Internet. Should Internet service providers and governments treat all data equally, or can they give preference to some content, sites, applications, users or devices based on defined criteria?

If you believe the pundits, network neutrality is about truth, justice and the unfettered right to watch YouTube videos. In reality, the network neutrality debate is a power and economic struggle between Internet service providers (i.e., mostly cable systems, telephone companies and wireless carriers) and those who provide content and services (e.g., movie and video services, search engines, social networks, web site hosts, email service providers and Internet businesses). Mediated by the U.S. government, the outcome of this battle has profound implications for all of us who consume content and use the Internet for work, recreation and social interaction. It will determine how much we pay for bandwidth, what services are available and how good they are.

To understand the net neutrality power struggle, it's instructive to consider an analogy with roads and traffic. Imagine a world where almost all roads were built, owned and operated by companies, rather than state and local governments. To recoup their construction costs and a profit, the companies charge tolls on all cars and trucks using the roads. To further complicate things, many of those road companies also operate their own fleets of cars and trucks, competing with their customers to deliver some goods and services.

The question is whether the road owners and operators can give priority to certain cars or trucks – their own or those of others -- based on how much the owners are willing to pay, the size of the fleets, and where the traffic goes. Superficially, the answer would seem to be obvious – yes, they should – until you realize that the goods and services on which individuals, businesses and governments all depend are being delivered by those cars and trucks. This conundrum is the balance of private enterprise and the public good.

One need look no further than the history of water companies, electric utilities, radio and television, and telephone companies to see the creative tension between regulation for the common good and stimulating free enterprise and supporting innovation. Many of us remember when the old AT&T (Ma Bell) was a regulated monopoly, with telephone rates and services subject to state and federal approval. The 1982 breakup of AT&T into a long distance company and a set of regional Bell operating companies (RBOCs) led to an explosion of wired and wireless services, competition from Sprint and MCI, and much lower long distance rates.

However, those old telephone regulations also supported the public good, ensuring that inexpensive telephone service was available everywhere, whether one lived in an isolated, rural area or a large city. Why? Because the regulations included cross-subsidies to support rural service. For all the excitement about high-speed Internet access, it is not universally available because the revenue in rural areas does not cover the costs of service. (Think about the toll road analogy again. Private sector toll roads are only economically attractive where there is high traffic.)

This brings us back to the network neutrality debate. The U.S. Federal Communications Commission (FCC) is proposing to treat the Internet, both wired and wireless, as a telecommunications service, using the same legal authority it uses to regulate other telecommunications services – radio, television, and telephony. Under the proposed rules, Internet broadband would be treated as a "common carrier," meaning the network must be open for all without discrimination. In addition, the FCC has proposed to refrain (forbear) from enforcing those regulatory provisions not relevant to modern broadband services. Nor would the proposal regulate rates or mandate service.

The political and economic battle is raging over how much regulation would be applied and how those regulations would affect the players. Some argue that any regulation will stifle innovation and private sector investment, and bring us a step closer to big brother rate regulation and tariffs – in the spirit of Ma Bell oversight.

I think that is highly unlikely, as the Internet is filled with competing services and companies. Rather, the network neutrality debate is about balancing innovation, private enterprise and the public good, while ensuring fairness and equal access. Simply put, it is a debate about appropriate checks and balances for a critical service. It should not be a political issue.

Internet access was once a luxury, but no longer. In a 21^st century knowledge economy, high-speed Internet services are the roads, waterways and rail lines of trade and commerce. They are the successors to telegraphy and telephony, coupling individuals and families across time and space. They are society's essential services, especially in a state like Iowa, where we are deeply dependent on communication for our future.

We need policies that enable universal access, while also ensuring Internet service providers have the flexibility and freedom to innovate and the economic returns to make that attractive. We also need mechanisms and technologies that support both top down and bottom up network deployment, allowing new market entrants and existing companies to expand coverage.

Remember, network neutrality is not just about streaming high definition movies to your television or mobile device, or buying goods and services via an e-commerce site. It's also about broadband access for education, electronic consultation with a medical specialist to avoid a repeat visit to the hospital during a winter storm, small businesses marketing products globally, and precision agriculture that lets Iowa farmers maximize crop yields. In short, network neutrality is complicated, but it matters to all of us.