But there are tens of thousands of Chinese characters, and a 5-by-7 grid was too small to make them legible. Chinese required a grid of 16 by 16 or larger—i.e., at least 32 bytes of memory (256 bits) per character. Were one to imagine a font containing 70,000 low-resolution Chinese characters, the total memory requirement would exceed two megabytes. Even a font containing only 8,000 of the most common Chinese characters would require approximately 256 kilobytes just to store the bitmaps. That was four times the total memory capacity of most off-the-shelf personal computers in the early 1980s.
As serious as these memory challenges were, the most taxing problems confronting low-res Chinese font production in the 1970s and 1980s were ones of aesthetics and design. Long before anyone sat down with a program like Gridmaster, the lion’s share of work took place off the computer, using pen, paper, and correction fluid.
Designers spent years trying to fashion bitmaps that fulfilled the low-memory requirements and preserved a modicum of calligraphic elegance. Among those who created this character set, whether by hand-drawing drafts of bitmaps for specific Chinese characters or digitizing them using Gridmaster, were Lily Huan-Ming Ling (凌焕銘) and Ellen Di Giovanni.
The core problem that designers faced was translating between two radically different ways of writing Chinese: the hand-drawn character, produced with pen or brush, and the bitmap glyph, produced with an array of pixels arranged on two axes. Designers had to decide how (and whether) they were going to try to re-create certain orthographic features of handwritten Chinese, such as entrance strokes, stroke tapering, and exit strokes.
In the case of the Sinotype III font, the process of designing and digitizing low-resolution Chinese bitmaps was thoroughly documented. One of the most fascinating archival sources from this period is a binder full of grids with hand-drawn hash marks all over them—sketches that would later be digitized into bitmaps for many thousands of Chinese characters. Each of these characters was carefully laid out and, in most cases, edited by Louis Rosenblum and GARF, using correction fluid to erase any “bits” the editor disagreed with. Over top of the initial set of green hash marks, then, a second set of red hash marks indicated the “final” draft. Only then did the work of data entry begin.
Given the sheer number of bitmaps that the team needed to design—at least 3,000 (and ideally many more) if the machine had any hopes of fulfilling consumers’ needs—one might assume that the designers looked for ways to streamline their work. One way they could have done this, for example, would have been to duplicate Chinese radicals—the base components of a character—when they appeared in roughly the same location, size, and orientation from one character to another. When producing the many dozens of common Chinese characters containing the “woman radical” (女), for example, the team at GARF could have (and, in theory, should have) created just one standard bitmap, and then replicated it within every character in which that radical appeared.
No such mechanistic decisions were made, however, as the archival materials show. On the contrary, Louis Rosenblum insisted that designers adjust each of these components—often in nearly imperceptible ways—to ensure they were in harmony with the overall character in which they appeared.
In the bitmaps for juan (娟, graceful) and mian (娩, to deliver), for example—each of which contains the woman radical—that radical has been changed ever so slightly. In the character juan, the middle section of the woman radical occupies a horizontal span of six pixels, as compared with five pixels in the character mian. At the same time, however, the bottom-right curve of the woman radical extends outward just one pixel further in the character mian, and in the character juan that stroke does not extend at all.
Across the entire font, this level of precision was the rule rather than the exception.
When we juxtapose the draft bitmap drawings against their final forms, we see that more changes have been made. In the draft version of luo (罗, collect, net), for example, the bottom-left stroke extends downward at a perfect 45° angle before tapering into the digitized version of an outstroke. In the final version, however, the curve has been “flattened,” beginning at 45° but then leveling out.
Despite the seemingly small space in which designers had to work, they had to make a staggering number of choices. And every one of these decisions affected every other decision they made for a specific character, since adding even one pixel often changed the overall horizontal and vertical balance.
The unforgiving size of the grid impinged upon the designers’ work in other, unexpected ways. We see this most clearly in the devilish problem of achieving symmetry. Symmetrical layouts—which abound in Chinese characters—were especially difficult to represent in low-resolution frameworks because, by the rules of mathematics, creating symmetry requires odd-sized spatial zones. Bitmap grids with even dimensions (such as the 16-by-16 grid) made symmetry impossible. GARF managed to achieve symmetry by, in many cases, using only a portion of the overall grid: just a 15-by-15 region within the overall 16-by-16 grid. This reduced the amount of usable space even further.
The story becomes even more complex when we begin to compare the bitmap fonts created by different companies or creators for different projects. Consider the water radical (氵) as it appeared in the Sinotype III font (below and on the right), as opposed to another early Chinese font created by H.C. Tien (on the left), a Chinese-American psychotherapist and entrepreneur who experimented with Chinese computing in the 1970s and 1980s.
As minor as the above examples might seem, each represented yet another decision (among thousands) that the GARF design team had to make, whether during the drafting or the digitization phase.
Low resolution did not stay “low” for long, of course. Computing advances gave rise to ever denser bitmaps, ever faster processing speeds, and ever diminishing costs for memory. In our current age of 4K resolution, retina displays, and more, it may be hard to appreciate the artistry—both aesthetic and technical—that went into the creation of early Chinese bitmap fonts, as limited as they were. But it was problem-solving like this that ultimately made computing, new media, and the internet accessible to one-sixth of the global population.
This AI could predict 10 years of scientific priorities—if we let it
The survey committee, which receives input from a host of smaller panels, takes into account a gargantuan amount of information to create research strategies. Although the Academies won’t release the committee’s final recommendation to NASA for a few more weeks, scientists are itching to know which of their questions will make it in, and which will be left out.
“The Decadal Survey really helps NASA decide how they’re going to lead the future of human discovery in space, so it’s really important that they’re well informed,” says Brant Robertson, a professor of astronomy and astrophysics at UC Santa Cruz.
One team of researchers wants to use artificial intelligence to make this process easier. Their proposal isn’t for a specific mission or line of questioning; rather, they say, their AI can help scientists make tough decisions about which other proposals to prioritize.
The idea is that by training an AI to spot research areas that are either growing or declining rapidly, the tool could make it easier for survey committees and panels to decide what should make the list.
“What we wanted was to have a system that would do a lot of the work that the Decadal Survey does, and let the scientists working on the Decadal Survey do what they will do best,” says Harley Thronson, a retired senior scientist at NASA’s Goddard Space Flight Center and lead author of the proposal.
Although members of each committee are chosen for their expertise in their respective fields, it’s impossible for every member to grasp the nuance of every scientific theme. The number of astrophysics publications increases by 5% every year, according to the authors. That’s a lot for anyone to process.
That’s where Thronson’s AI comes in.
It took just over a year to build, but eventually, Thronson’s team was able to train it on more than 400,000 pieces of research published in the decade leading up to the Astro2010 survey. They were also able to teach the AI to sift through thousands of abstracts to identify both low- and high-impact areas from two- and three-word topic phrases like “planetary system” or “extrasolar planet.”
According to the researchers’ white paper, the AI successfully “backcasted” six popular research themes of the last 10 years, including a meteoric rise in exoplanet research and observation of galaxies.
“One of the challenging aspects of artificial intelligence is that they sometimes will predict, or come up with, or analyze things that are completely surprising to the humans,” says Thronson. “And we saw this a lot.”
Thronson and his collaborators think the steering committee should use their AI to help review and summarize the vast amounts of text the panel must sift through, leaving human experts to make the final call.
Their research isn’t the first to try to use AI to analyze and shape scientific literature. Other AIs have already been used to help scientists peer-review their colleagues’ work.
But could it be trusted with a task as important and influential as the Decadal Survey?
Securing the energy revolution and IoT future
In early 2021, Americans living on the East Coast got a sharp lesson on the growing importance of cybersecurity in the energy industry. A ransomware attack hit the company that operates the Colonial Pipeline—the major infrastructure artery that carries almost half of all liquid fuels from the Gulf Coast to the eastern United States. Knowing that at least some of their computer systems had been compromised, and unable to be certain about the extent of their problems, the company was forced to resort to a brute-force solution: shut down the whole pipeline.
The interruption of fuel delivery had huge consequences. Fuel prices immediately spiked. The President of the United States got involved, trying to assure panicked consumers and businesses that fuel would become available soon. Five days and untold millions of dollars in economic damage later, the company paid a $4.4 million ransom and restored its operations.
It would be a mistake to see this incident as the story of a single pipeline. Across the energy sector, more and more of the physical equipment that makes and moves fuel and electricity across the country and around the world relies on digitally controlled, networked equipment. Systems designed and engineered for analogue operations have been retrofitted. The new wave of low-emissions technologies—from solar to wind to combined-cycle turbines—are inherently digital tech, using automated controls to squeeze every efficiency from their respective energy sources.
Meanwhile, the covid-19 crisis has accelerated a separate trend toward remote operation and ever more sophisticated automation. A huge number of workers have moved from reading dials at a plant to reading screens from their couch. Powerful tools to change how power is made and routed can now be altered by anyone who knows how to log in.
These changes are great news—the world gets more energy, lower emissions, and lower prices. But these changes also highlight the kinds of vulnerabilities that brought the Colonial Pipeline to an abrupt halt. The same tools that make legitimate energy-sector workers more powerful become dangerous when hijacked by hackers. For example, hard-to-replace equipment can be given commands to shake itself to bits, putting chunks of a national grid out of commission for months at a stretch.
For many nation-states, the ability to push a button and sow chaos in a rival state’s economy is highly desirable. And the more energy infrastructure becomes hyperconnected and digitally managed, the more targets offer exactly that opportunity. It’s not surprising, then, that an increasing share of cyberattacks seen in the energy sector have shifted from targeting information technologies (IT) to targeting operating technologies (OT)—the equipment that directly controls physical plant operations.
To stay on top of the challenge, chief information security officers (CISOs) and their security operations centers (SOCs) will have to update their approaches. Defending operating technologies calls for different strategies—and a distinct knowledge base—than defending information technologies. For starters, defenders need to understand the operating status and tolerances of their assets—a command to push steam through a turbine works well when the turbine is warm, but can break it when the turbine is cold. Identical commands could be legitimate or malicious, depending on context.
Even collecting the contextual data needed for threat monitoring and detection is a logistical and technical nightmare. Typical energy systems are composed of equipment from several manufacturers, installed and retrofitted over decades. Only the most modern layers were built with cybersecurity as a design constraint, and almost none of the machine languages used were ever meant to be compatible.
For most companies, the current state of cybersecurity maturity leaves much to be desired. Near-omniscient views into IT systems are paired with big OT blind spots. Data lakes swell with carefully collected outputs that can’t be combined into a coherent, comprehensive picture of operational status. Analysts burn out under alert fatigue while trying to manually sort benign alerts from consequential events. Many companies can’t even produce a comprehensive list of all the digital assets legitimately connected to their networks.
In other words, the ongoing energy revolution is a dream for efficiency—and a nightmare for security.
Securing the energy revolution calls for new solutions equally capable of identifying and acting on threats from both physical and digital worlds. Security operations centers will need to bring together IT and OT information flows, creating a unified threat stream. Given the scale of data flows, automation will need to play a role in applying operational knowledge to alert generation—is this command consistent with business as usual, or does context show it’s suspicious? Analysts will need broad, deep access to contextual information. And defenses will need to grow and adapt as threats evolve and businesses add or retire assets.
This month, Siemens Energy unveiled a monitoring and detection platform aimed at resolving the core technical and capability challenges for CISOs tasked with defending critical infrastructure. Siemens Energy engineers have done the legwork needed to automate a unified threat stream, allowing their offering, Eos.ii, to serve as a fusion SOC that’s capable of unleashing the power of artificial intelligence on the challenge of monitoring energy infrastructure.
AI-based solutions answer the dual need for adaptability and persistent vigilance. Machine learning algorithms trawling huge volumes of operational data can learn the expected relationships between variables, recognizing patterns invisible to human eyes and highlighting anomalies for human investigation. Because machine learning can be trained on real-world data, it can learn the unique characteristics of each production site, and can be iteratively trained to distinguish benign and consequential anomalies. Analysts can then tune alerts to watch for specific threats or ignore known sources of noise.
Extending monitoring and detection into the OT space makes it harder for attackers to hide—even when unique, zero-day attacks are deployed. In addition to examining traditional signals like signature-based detection or network traffic spikes, analysts can now observe the effects that new inputs have on real-world equipment. Cleverly disguised malware would still raise red flags by creating operational anomalies. In practice, analysts using the AI-based systems have found that their Eos.ii detection engine was sensitive enough to predictively identify maintenance needs—for example, when a bearing begins to wear out and the ratio of steam in to power out begins to drift.
Done right, monitoring and detection that spans both IT and OT should leave intruders exposed. Analysts investigating alerts can trace user histories to determine the source of anomalies, and then roll forward to see what else was changed in a similar timeframe or by the same user. For energy companies, increased precision translates to dramatically reduced risk – if they can determine the scope of an intrusion, and identify which specific systems were compromised, they gain options for surgical responses that fix the problem with minimal collateral damage—say, shutting down a single branch office and two pumping stations instead of a whole pipeline.
As energy systems continue their trend toward hyperconnectivity and pervasive digital controls, one thing is clear: a given company’s ability to provide reliable service will depend more and more on their ability to create and sustain strong, precise cyber defenses. AI-based monitoring and detection offers a promising start.
To learn more about Siemens Energy’s new AI-based monitoring and detection platform, check out their recent white paper on Eos.ii.
Learn more about Siemens Energy cybersecurity at Siemens Energy Cybersecurity.
This content was produced by Siemens Energy. It was not written by MIT Technology Review’s editorial staff.
The US is about to kick-start its controversial covid booster campaign
Disagreements: Booster shots have been controversial. A group of top scientists, including experts at the FDA and WHO, published a review in The Lancet on Monday arguing that booster shots are unnecessary since vaccines are still very effective at preventing severe disease and death. Furthermore, they say, vaccine supplies could save more lives if they’re used for unvaccinated people rather than as boosters for the vaccinated. That’s why the WHO has been pleading with rich countries to stop handing them out until more of the world is vaccinated.
Unequal distribution: The US joins the UK, the UAE, France, Germany, and Israel, which have also launched booster programs. In the UK, for example, a rollout of booster shots to all over-50s is about to begin after officials gave the green light last week. Meanwhile, less than 4% of Africa’s population is fully vaccinated, compared with 70% of adults in the EU. In the US, it’s 55%, a figure that has stubbornly failed to significantly budge in recent weeks. Earlier this week, President Biden announced that the US would buy a further 500 million doses of vaccine to distribute to other parts of the world, bringing its total commitment to more than 1 billion.
Scramble: Millions of Americans are likely to try to get a third shot. A YouGov poll this summer found that three in five vaccinated Americans will get one if it’s available. Given the chaotic nature of the US vaccine rollout, it will be hard to prevent people from gaming the system to get a third shot even if they aren’t technically eligible.