Using chip memory more efficiently

For decades, computer chips have increased efficiency by using “caches,” small, local memory banks that store frequently used data and cut down on time- and energy-consuming communication with off-chip memory.

Today’s chips generally have three or even four different levels of cache, each of which is more capacious but slower than the last. The sizes of the caches represent a compromise between the needs of different kinds of programs, but it’s rare that they’re exactly suited to any one program.

Researchers at MIT’s Computer Science and Artificial Intelligence Laboratory have designed a system that reallocates cache access on the fly, to create new “cache hierarchies” tailored to the needs of particular programs.

The researchers tested their system on a simulation of a chip with 36 cores, or processing units. They found that, compared to its best-performing predecessors, the system increased processing speed by 20 to 30 percent while reducing energy consumption by 30 to 85 percent.

“What you would like is to take these distributed physical memory resources and build application-specific hierarchies that maximize the performance for your particular application,” says Daniel Sanchez, an assistant professor in the Department of Electrical Engineering and Computer Science (EECS), whose group developed the new system.

“And that depends on many things in the application. What’s the size of the data it accesses? Does it have hierarchical reuse, so that it would benefit from a hierarchy of progressively larger memories? Or is it scanning through a data structure, so we’d be better off having a single but very large level? How often does it access data? How much would its performance suffer if we just let data drop to main memory? There are all these different tradeoffs.”

Sanchez and his coauthors — Po-An Tsai, a graduate student in EECS at MIT, and Nathan Beckmann, who was an MIT graduate student when the work was done and is now an assistant professor of computer science at Carnegie Mellon University — presented the new system, dubbed Jenga, at the International Symposium on Computer Architecture last week.

Staying local

For the past 10 years or so, improvements in computer chips’ processing power have come from the addition of more cores. The chips in most of today’s desktop computers have four cores, but several major chipmakers have announced plans to move to six cores in the next year or so, and 16-core processors are not uncommon in high-end servers. Most industry watchers assume that the core count will continue to climb.

Each core in a multicore chip usually has two levels of private cache. All the cores share a third cache, which is actually broken up into discrete memory banks scattered around the chip. Some new chips also include a so-called DRAM cache, which is etched into a second chip that is mounted on top of the first.

For a given core, accessing the nearest memory bank of the shared cache is more efficient than accessing more distant cores. Unlike today’s cache management systems, Jenga distinguishes between the physical locations of the separate memory banks that make up the shared cache. For each core, Jenga knows how long it would take to retrieve information from any on-chip memory bank, a measure known as “latency.”

Jenga builds on an earlier system from Sanchez’s group, called Jigsaw, which also allocated cache access on the fly. But Jigsaw didn’t build cache hierarchies, which makes the allocation problem much more complex.

For every task running on every core, Jigsaw had to calculate a latency-space curve, which indicated how much latency the core could expect with caches of what size. It then had to aggregate all those curves to find a space allocation that minimized latency for the chip as a whole.

Curves to surfaces

But Jenga has to evaluate the tradeoff between latency and space for two layers of cache simultaneously, which turns the two-dimensional latency-space curve into a three-dimensional surface. Fortunately, that surface turns out to be fairly smooth: It may undulate, but it usually won’t have sudden, narrow spikes and dips.

That means that sampling points on the surface will give a pretty good sense of what the surface as a whole looks like. The researchers developed a clever sampling algorithm tailored to the problem of cache allocation, which systematically increases the distances between sampled points. “The insight here is that caches with similar capacities — say, 100 megabytes and 101 megabytes — usually have similar performance,” Tsai says. “So a geometrically increased sequence captures the full picture quite well.”

Once it has deduced the shape of the surface, Jenga finds the path across it that minimizes latency. Then it extracts the component of that path contributed by the first level of cache, which is a 2-D curve. At that point, it can reuse Jigsaw’s space-allocation machinery.

In experiments, the researchers found that this approach yielded an aggregate space allocation that was, on average, within 1 percent of that produced by a full-blown analysis of the 3-D surface, which would be prohibitively time consuming. Adopting the computational short cut enables Jenga to update its memory allocations every 100 milliseconds, to accommodate changes in programs’ memory-access patterns.

End run

Jenga also features a data-placement procedure motivated by the increasing popularity of DRAM cache. Because they’re close to the cores accessing them, most caches have virtually no bandwidth restrictions: They can deliver and receive as much data as a core needs. But sending data longer distances requires more energy, and since DRAM caches are off-chip, they have lower data rates.

If multiple cores are retrieving data from the same DRAM cache, this can cause bottlenecks that introduce new latencies. So after Jenga has come up with a set of cache assignments, cores don’t simply dump all their data into the nearest available memory bank. Instead, Jenga parcels out the data a little at a time, then estimates the effect on bandwidth consumption and latency. Thus, even within the 100-millisecond intervals between chip-wide cache re-allocations, Jenga adjusts the priorities that each core gives to the memory banks allocated to it.

“There’s been a lot of work over the years on the right way to design a cache hierarchy,” says David Wood, a professor of computer science at the University of Wisconsin at Madison. “There have been a number of previous schemes that tried to do some kind of dynamic creation of the hierarchy. Jenga is different in that it really uses the software to try to characterize what the workload is and then do an optimal allocation of the resources between the competing processes. And that, I think, is fundamentally more powerful than what people have been doing before. That’s why I think it’s really interesting.”

de MIT News http://ift.tt/2tUCsDl

MIT convenes ad hoc task force on open access to Institute’s research

MIT’s provost, in consultation with the vice president for research, the chair of the faculty, and the director of the libraries, has appointed an ad hoc task force on open access to MIT’s research. Convening the task force was one of the 10 recommendations presented in the preliminary report of the Future of Libraries Task Force.

The open access task force, chaired by Class of 1922 Professor of Electrical Engineering and Computer Science Hal Abelson and Director of Libraries Chris Bourg, will lead an Institute-wide discussion of ways in which current MIT open access policies and practices might be updated or revised to further the Institute’s mission of disseminating the fruits of its research and scholarship as widely as possible.

“To solve the world’s toughest challenges, we must lower the barriers to knowledge,” says Maria Zuber, vice president for research. “We want to share MIT’s research as widely and openly as we can, not only because it’s in line with our values but because it will accelerate the science and the scholarship that can lead us to a better world. I look forward to seeing the Institute strengthen its leadership position in open access through this task force’s work.”

Adopted in 2009, the MIT Faculty Open Access Policy allows MIT authors to legally hold onto rights in their scholarly articles, including the right to share them widely. It was one of the first and most far-reaching initiatives of its kind in the United States. MIT remains a leader in open access, with 44 percent of faculty journal articles published since the adoption of the policy freely available to the world. In April of this year, the Institute announced a new policy under which all MIT authors — including students, postdocs, and staff — can opt in to an open access license.

The preliminary report of the Institute-wide Future of Libraries Task Force, released in October 2016, acknowledged that while MIT’s open access policy and implementation are widely seen as a successful model, “the fact remains that most of MIT’s scholarship remains unavailable for open dissemination. … The gap in coverage not only represents a loss in access for MIT’s global community of stakeholders, it also ensures that MIT’s full contribution to the scholarly record cannot be comprehensively assessed or computationally analyzed.”

The task force’s activities will include reviewing MIT’s open access activities to date, as well as those of sister institutions and other organizations, and working with faculty and administration in MIT’s departments, labs, centers, and other units such as MITx, MIT Press, and the Office for Digital Learning, to discuss these initiatives and opportunities for enhancing them.

The members of the task force are:

• Hal Abelson (co-chair), Class of 1922 Professor in theDepartment of Electrical Engineering and Computer Science;
• Chris Bourg (co-chair), director of libraries;
• Peter Bebergal, technology licensing officer in the Technology Licensing Office;
• Robert Bond, associate head of the Intelligence, Surveillance, and Reconnaissance and Tactical Systems Division at MIT Lincoln Laboratory;
• Herng Yi Cheng, undergraduate in the Department of Mathematics;
• Isaac Chuang, professor in the Department of Electrical Engineering and Computer Science and senior associate dean of digital learning;
• Christopher Cummins, the Henry Dreyfus Professor of Chemistry;
• Deborah Fitzgerald, professor in the Program in Science, Technology, and Society;
• Mark Jarzombek, professor in the Department of Architecture;
• Nick Lindsay, journals director at the MIT Press;
• Jack Reid, graduate student in the Technology and Policy Program and the Department of Aeronautics and Astronautics;
• Karen Shirer, director of research development in the Office of the Vice President for Research;
• Bernhardt Trout, professor in the Department of Chemical Engineering;
• Eric von Hippel, the T. Wilson (1953) Professor in Management; and
• Jay Wilcoxson, counsel in the Office of the General Counsel.

The task force will prepare recommendations to the administration, and as appropriate, to the faculty, for new and strengthened open access initiatives, together with possible changes to MIT policies, and will work with the administration to develop implementation plans.

de MIT News http://ift.tt/2sxCldp

Why do some neighborhoods improve?

Four years ago, researchers at MIT’s Media Lab developed a computer vision system that can analyze street-level photos taken in urban neighborhoods in order to gauge how safe the neighborhoods would appear to human observers.

Now, in an attempt to identify factors that predict urban change, the MIT team and colleagues at Harvard University have used the system to quantify the physical improvement or deterioration of neighborhoods in five American cities.

In work reported today in the Proceedings of the National Academy of Sciences, the system compared 1.6 million pairs of photos taken seven years apart. The researchers used the results of those comparisons to test several hypotheses popular in the social sciences about the causes of urban revitalization. They find that density of highly educated residents, proximity to central business districts and other physically attractive neighborhoods, and the initial safety score assigned by the system all correlate strongly with improvements in physical condition.

Perhaps more illuminating, however, are the factors that turn out not to predict change. Raw income levels do not, and neither do housing prices or neighborhoods’ ethnic makeup.

“So it’s not an income story — it’s not that there are rich people there, and they happen to be more educated,” says César Hidalgo, the Asahi Broadcasting Corporation Associate Professor of Media Arts and Sciences and senior author on the paper. “It appears to be more of a skill story.”

Tipping points

“That’s the first theory we found support for,” adds Nikhil Naik, a postdoc at MIT’s Abdul Latif Jameel Poverty Action Lab and first author on the new paper. “And the second theory was the the so-called tipping theory, which says that neighborhoods that are already doing well will continue to do better, and neighborhoods that are not doing well will not improve as much.”

While the researchers found that, on average, higher initial safety scores did indeed translate to larger score increases over time, the relationship was linear: A neighborhood with twice the initial score of another would see about twice as much improvement. This contradicts the predictions of some theorists, who have argued that past some “tipping point,” improvements in a neighborhood’s quality should begin to accelerate.

The researchers also tested the hypothesis that neighborhoods tend to be revitalized when their buildings have decayed enough to require replacement or renovation. But they found little correlation between the average age of a neighborhood’s buildings and its degree of physical improvement.

Joining Naik and Hidalgo on the paper are Ramesh Raskar, an associate professor of media arts and sciences, who, with Hidalgo, supervised Naik’s PhD thesis in the Media Lab, and two Harvard professors: Scott Kominers, an associate professor of entrepreneurial management at the Harvard Business School, and Edward Glaeser, an economics professor.

Noisy signals

The system that assigned the safety ratings was a machine-learning system, which had been trained on hundreds of thousands of examples in which human volunteers had rated the relative safety of streetscapes depicted in pairs of images. In the new study, the system compared images associated with the same geographic coordinates from Google’s Street View visualization tool, but captured seven years apart.

Those images had to be preprocessed, however, to ensure that the system’s inferred changes in perceived safety were reliable. For instance, previous work from Hidalgo’s group suggested that prevalence of green spaces was one of the criteria that human volunteers used in assessing safety. But if the earlier of a pair of images was captured in summer, and the later was captured in winter, the machine-learning system might be fooled into thinking that the neighborhood had lost green space.

Similarly, the prevalence of buildings with street-facing windows also appeared to increase neighborhoods’ safety scores. But if a panel truck in the foreground of an image obscured three floors’ worth of windows in the building behind it, the system might assign the image an artificially low score.

So the researchers used a computer-vision technique called semantic segmentation to categorize every pixel of every one of the 1.6 million images in their data set according to the object that comprised it. If something like a truck or a pedestrian constituted too much of an image, the system rejected the image and instead compared images associated with different coordinates on the same block. Similarly, in assessing the perceived safety of a streetscape, the system ignored those parts of the image, such as trees and skies, that were too susceptible to seasonal vicissitudes.

To validate the system’s analyses, the researchers also presented 15,000 randomly selected pairs of images from their data set to reviewers recruited through Amazon’s Mechanical Turk crowdsourcing platform, who were asked to assess the relative safety of the neighborhoods depicted. The reviewers’ assessments coincided with the computer system’s 72 percent of the time. But most of the disagreements centered on pairs of images with little change in safety scores; in those borderline cases, any two humans might disagree, too.

“I think this is really interesting and vsionary work by top-of-the-line researchers,” says Julia Lane, a professor at New York University’s Center for Urban Science and Progress. “I hope to see more work like this. The combination of high-quality measurement, analysis, and thoughtful attention to what is missing is the future of measurement.” 

de MIT News http://ift.tt/2uwjp0b

Lucas Mason-Brown named 2017 Echoing Green Fellow

MIT mathematics graduate student Lucas Mason-Brown has been named one of 35 Echoing Green Fellows for his work with Data for Black Lives (D4BL), an organization he recently co-founded to mobilize scientists to use data science to fight racial bias in real estate, finance, criminal justice, and other areas.

Historically, data science has been used in ways that disproportionately and negatively affect black communities. Big data and algorithms have been instrumental in predatory lending practices, predictive policing, and redlining, the practice of denying services such as banking or insurance to residents of a specific area based on its racial or ethnic composition.

D4BL aims to turn data science tools such as statistical modeling, data visualization, and crowdsourcing into instruments for fighting bias, building progressive movements, and promoting civic engagement.

Until last summer, D4BL was just a Twitter account with a few dozen followers until Mason-Brown and his friend and co-founder Yeshimabeit Milner put their ideas into action with a third friend, Max Clermont, who works in public health.  

Now, with $90,000 of seed-funding that the Echoing Green Fellowship will provide over the next two years, D4BL will hire Milner as full-time executive director. Echoing Green is a nonprofit that has kickstarted both nonprofit and for-profit social entrepreneurs in more than 60 countries since 1987. Other fellowship benefits include health insurance, professional development, networking opportunities, technical support, pro bono partnerships, and a dedicated Echoing Green portfolio manager to help grow their organization. D4BL joins a community of social impact leaders that include Teach for America, City Year, One Acre Fund, and SKS Microfinance.

Public launch

D4BL will also host a conference at the MIT Media Lab in November, an event that will also serve as D4BL’s public launch. The conference is expected to gather more than 200 activists, organizers, data scientists, computer programmers, and public officials.

Conference speakers from MIT will include President L. Rafael Reif, Dean of the School of Humanities and Social Science Melissa Nobles, and Media Lab Director’s Fellows Adam Foss and Julia Angwin, as well as Cathy O’Neil, MIT instructor and author of "Weapons of Math Destruction." Events will include a discussion of the mathematics of gerrymandering congressional districts led by the Metric Geometry and Gerrymandering Group, and a hackathon that will encourage mathematicians and data scientists to work with activists and organizers on pressing racial justice issues. 

“The issues focused around people of color in this country deeply resonate with me,” said Isaiah Borne, a rising senior in chemical engineering, conference organizer, and political action co-chair of the MIT Black Students’ Union. “The D4BL conference is a unique opportunity for me — and anyone who's involved — to look at these social issues from a different perspective and create real, meaningful change in our community.”

The conference has also received support from MIT Vice President Kirk Kolenbrander, Vice President of Student Life Suzy Nelson, Chancellor Cynthia Barnhart, Institute for Data Systems and Society Director Munther Dahleh, Media Lab Director Joi Ito, and physics Professor Ed Bertschinger, who is co-chair of both the MIT Committee on Race and Diversity and the Faculty Advisory Committee of the Office of Minority Education.   

“We have been overwhelmed by the amount of support and interest we have received at MIT and beyond,” said Mason-Brown. “There is a real thirst and a real need for this kind of work.”

A student of representation theory

A native of Belmont, Massachusetts, Mason-Brown studied math and philosophy at Brown University, received his MS in mathematics from Trinity College in Dublin, and taught seventh grade math and science for a year at the Edward Brooke School in Roslindale, Massachusetts. He is now in his second year at MIT, where he studies representation theory — the study of abstract symmetries — with his “mathematical hero,” Professor David Vogan.

Mason-Brown's research may not be directly related to his work for D4BL, but he is able to make time for both, he says.

“I think for me the impetus is simple: When you discover that something you care about deeply has been used, intentionally or unintentionally, to harm communities across the country — as big data and algorithms certainly have — you have no choice but to stand up and act.”

de MIT News http://ift.tt/2tv3aA5

Evaluating approaches to agricultural development

As the old saying goes, teaching someone to fish is far more helpful than just giving them a fish. Now, research from WorldFish and MIT takes that adage a step further: Better yet, the study found, is working with the fishermen to help develop better fishing methods.

Involving local people in figuring out how to improve their farming and fishing methods provides more lasting and widespread benefits than just introducing new technologies or methods, the researchers showed. The findings are described in the journal Agricultural Systems, in a paper by Boru Douthwaite of the research funding agency WorldFish, based in Malaysia, and Elizabeth Hoffecker, lead researcher at the International Development Innovation Network (IDIN), based at the MIT D-Lab.

Considerable research over the last few decades has shown that bringing about improvements in agricultural systems is a highly complex challenge, with many interrelationships and feedbacks determining how well new methods and devices take hold or provide a real improvement. Yet government agencies as well as research and nonprofit organizations still mostly evaluate the success of their programs using simple metrics that overlook much of this complexity, Hoffecker says.

For three decades, Douthwaite has been studying how these programs work in practice. He says he has often observed a disconnect between the measures agencies use to decide whether a program is working, versus the real effects he saw in some of the communities involved.

For this study, the researchers focused on two quite different examples that help to illustrate these disparities: fishing on lakes and rivers in Zambia, and growing a fiber crop called abaca in the Philippines.

In Zambia, a landlocked country in southern Africa, refrigeration facilities and ice are scarce, and nearly one-third of the fish caught there is lost to spoilage before it ever reaches the market. Many of the fish are currently dried or smoked, but they are vulnerable to insects and rodents during the drying process, and they become brittle and subject to damage during transportation. Reducing such spoilage could both provide financial benefits for the economically struggling fishing community and help alleviate food shortages for consumers.

The Zambian fisheries were facing two related issues, Hoffecker says: “The narrow challenge was to come up with a way to prevent fish from spoiling.” But in addressing that challenge, it became apparent that “there was a much bigger challenge, which was overfishing.” Though many communities in the region were facing these same challenges, “some of the stakeholders were not working together” to address them, she says. If people had tried to get these groups to work together on the bigger challenge right at the start, she says, “it probably would have failed,” because there was so much mistrust between the different communities.

But instead, she said, “they started out working on this technical challenge,” of reducing spoilage, “which built relationships that allowed them to tackle the bigger challenge.” The participatory research process included meetings of different stakeholders including government officials, non-governmental organizations, researchers, and residents, which were followed by village-level workshops in 10 communities. This resulted in establishing three ongoing working groups to tackle different aspects of the issue: fisheries co-management, establishing cooperatives and other economic associations, and postharvest processing. Among other solutions, the group decided to introduce salting of fish as an improved preservation method.

The overall process led to four significant outcomes, Douthwaite says — none of which had been planned or anticipated initially and thus might have been missed in an evaluation based just on meeting initial, stated goals. The four outcomes consisted of developing a locally sourced fish-processing method (the salting), developing a value chain for the salted fish from harvest to market, creating working groups that could continue to evaluate and improve innovations in the fishery, and improving relationships among the different groups involved, from the fishermen to the government agencies to the traders and buyers. In the end, this led to a growing consensus about the need for measures to prevent overfishing.

In the other case studied, the Philippine abaca farmers had been facing a virus that threatened to greatly diminish their harvests of the widely used fiber plant, which is the nation’s primary source of cordage and paper. With some regions experiencing a 90 percent decline in harvests, the government’s initial strategy was to eradicate all the infected plants to curtail the virus’ spread. But farmers were wary of efforts to destroy the plants they relied on, especially when there was miscommunication about what exactly was being done.

So when a new, virus-resistant variety of the plant was developed, the local farmers’ weren’t willing to make the switch, as they considered the new varieties inferior for fiber-making.

Instead of just pressing the farmers to change, the team used a different approach, “enlisting the farmers in a process of experimentation,” as Hoffecker describes it. Several hybrid varieties were developed, and the local farmers tested them in their fields. “Because they were involved in the process, they were much more receptive to the results,” she says. In fact, many of them came up with their own suggestions for furthering the research, including testing local varieties that seemed to be naturally resistant and trying plantings on different kinds of soils and slopes.

They not only came up with an acceptable resistant variety, but when it turned out there were not enough seeds available, the farmers developed their own strategy for sharing the seeds, requiring those who got the initial seedling allotments to pay back new seedlings into the system for others to plant.

Many development organizations are well aware of these kinds of complexities and of the need for more community involvement and fewer “top-down” aid solutions, Hoffecker says. A problem, though, is that the metrics and results-assessment frameworks used to measure success often leave no room for complex, emergent outcomes. Instead, they typically focus on measuring the extent to which various solutions — such as a new crop variety, tool, or farming method — are adopted, equating scale of use with success.

Outcomes associated with how the solutions were developed, such as the creation of greater cooperation among communities or stakeholder groups, and the instilling of local empowerment and problem-solving abilities and motivation, are much harder to measure and typically left to anecdotes rather than rigorously assessed, she says.

This research was designed to help provide a basis for new ways to assess the success of programs that are working towards these types of outcomes. “It’s a first step in developing such a model and encouraging others to develop such models,” Douthwaite says.

And such assessments are essential, Hoffecker says, for making development interventions more effective and lasting. Some worthy projects, she says, are “not getting funded, because the results are not understood by the donors. Some projects are producing important outcomes, but they’re not being seen and appreciated.” Hopefully, she says, this new study can begin to address that need.

The fieldwork for this research was funded by WorldFish, which is part of the Consultative Group on International Agricultural Research, a World Bank agency that promotes research on sustainable development of agriculture. 

de MIT News http://ift.tt/2uvoV39

A simple solution for terrible traffic

Cities plagued with terrible traffic problems may be overlooking a simple, low-cost solution: High-occupancy vehicle (HOV) policies that encourage carpooling can reduce traffic drastically, according to a new study co-authored by MIT economists.

The results show that in Jakarta, Indonesia, travel delays became 46 percent worse during the morning rush hour and 87 percent worse during the evening rush hour, after an HOV policy requiring three or more passengers in a car was discontinued on important city center roads.

“Eliminating high-occupancy vehicle restrictions led to substantially worse traffic,” says Ben Olken, a professor of economics at MIT and co-author of a new paper detailing the study. “That’s not shocking, but the magnitudes are just enormous.”

Moreover, when the HOV policy was ended, traffic suddenly became significantly worse on surrounding roads as well. Instead of siphoning more traffic onto the central roads, the policy change made congestion worse all over.

“HOV policies on central roads were making traffic everywhere better, both during the middle of the day and on these other roads during rush hour,” Olken observes. “That I think is a really striking result.”

The paper, “Citywide effects of high-occupancy vehicle restrictions: Evidence from ‘three-in-one’ in Jakarta,” is being published today in the journal Science. The authors are Olken; Gabriel Kreindler, a doctoral candidate in development economics in the MIT Department of Economics; and Rema Hanna, the Jeffrey Cheah Professor of South-East Asia Studies at the Harvard Kennedy School. Olken is the corresponding author.

Drivers and jockeys

In an effort to reduce its notoriously bad traffic problems, Jakarta installed its HOV regulations in 1992, using a “three-in-one” policy that required three passengers in each vehicle on some major roads, between 7 and 10 a.m. and between 4:30 and 7 p.m. However, it suddenly scrapped the policy in 2016 — first for a week, then for a month, and then permanently.

The HOV system had drawn its share of critics locally, in part due to the presence of “jockeys” on the roads — people charging a small fee to drivers who would let them ride in their cars, bringing vehicles up to the three-person requirement. Commuters would typically pick up jockeys at the edges of the HOV area. This led onlookers to question whether the HOV policy really was reducing the number of vehicles effectively, since it seemed to limit the amount of ride-sharing by actual commuters.

“For a long time, people were skeptical of the policy because of the existence of these jockeys,” says Kreindler.

When the Jakarta government suddenly announced the policy change in late March 2016, however, it gave the researchers an ideal opportunity to conduct a before-and-after natural experiment. The team queried Google Maps data starting in late March, before the policy went into effect, and then continued monitoring its impact into June 2016.

“The key thing we did is to start collecting traffic data immediately,” Hanna explains. “Within 48 hours of the policy announcement, we were regularly having our computers check Google Maps every 10 minutes to check current traffic speeds on several roads in Jakarta. ... By starting so quickly we were able to capture real-time traffic conditions while the HOV policy was still in effect. We then compared the changes in traffic before and after the policy change.”

All told, the impact of changing the HOV policy was highly significant. After the HOV policy was abandoned, the average speed of Jakarta’s rush hour traffic declined from about 17 to 12 miles per hour in the mornings, and from about 13 to 7 miles per hour in the evenings. By comparison, people usually walk at around 3 miles per hour.

As the researchers acknowledge, the precise mechanism for the spillover effect of eliminating HOV lanes — the fact that traffic got worse on surrounding roads — could use further investigation. After all, intuitively, it might seem that discarding the HOV policy on major streets would lessen the amount of traffic on nearby roads.

The scholars suggest a few potential reasons for this larger effect. The most direct is that the HOV-lane policy, when in effect, reduced the overall number of cars on the roads by encouraging car-sharing — and the ensuing, post-HOV conditions simply increased overall traffic levels everywhere.

Another possibility is that ending the HOV policy may have led to “hypercongestion,” in which traffic clogs became so bad that the total volume of vehicles on the major arteries dropped, such that more people drove on the surrounding streets, leading to more tie-ups there too. Still another hypothesis would be that there was a slightly less dramatic but still-consequential spillover of vehicles from Jakarta’s central business district to feeder streets.

Whatever the case, Olken says, the empirical effect is genuine.

“The point of this paper is to show that in a real-world natural experiment, this is what happens,” Olken says.  
 
Attention: Urban planners

Kreindler, who is planning further traffic studies in other cities, notes that the sudden change in Jakarta’s HOV policies gives more weight to the findings, since it reduces the possibility that other factors were at work.

“The fact that it was unexpected means that the changes we’re looking at, before and after, are really due to the policy and not some other development that would also affect how much people would want to drive,” Kreindler says.

To be sure, the scholars note, all cities are different, and the results of similar changes might vary in magnitude elsewhere, depending on many factors, from the format of the urban area to the amount of public transit available.

Still, the result seems directly of interest to urban planners and policymakers, especially because of the low-cost nature of implementing HOV policies, whether on entire streets or certain lanes of large streets. All a city really needs, after all, are some signs, some paint for markings, and some enforcement of the policy.

“I don’t think we should necessarily take the result and wildly apply it everywhere, but [given] the kind of really serious congestion problems Jakarta has, it suggests this is a policy measure that has the potential to work,” Olken says.  

de MIT News http://ift.tt/2tNbluh

Study: Preschoolers learn from math games — to a point

What is the best way to help poor schoolchildren succeed at math? A study co-authored by researchers at MIT, Harvard University, and New York University now sheds light on the ways preschool activities may — or may not — help children develop cognitive skills.

The study, based on an experiment in Delhi, India, engaged preschool children in math games intended to help them grasp concepts of number and geometry, and in social games intended to help them cooperate and learn together.

The results contained an unexpected wrinkle. Children participating in the math games did retain a superior ability to grasp those concepts more than a year later, compared to children who either played only the social games or did not participate. However, the exercises did not lead to better results later, when the children entered a formal classroom setting.

“It’s very clear you have a significant improvement in the math skills” used in the games, says Esther Duflo, the Abdul Latif Jameel Professor of Poverty Alleviation and Development Economics at MIT and co-author of the study. “We find that the gains are persistent … which I think is quite striking.”

However, she adds, by the time the children in the study were learning formal math concepts in primary school, such as specific number symbols, the preschool intervention did not affect learning outcomes.

“All the kids [in primary school] had learned, but they had learned [those concepts] equally,” says Duflo, who is a co-founder of MIT’s Abdul Latif Jameel Poverty Action Lab (J-PAL), which conducts field experiments, often in education, around the globe.

A paper detailing the results of the study, “Cognitive science in the field: A preschool intervention durably enhances intuitive but not formal mathematics,” is being published today in the journal Science.

The authors are Duflo; Moira R. Dillon, an assistant professor in New York University’s Department of Psychology; Harini Kannan, a postdoc at J-PAL South Asia; Joshua T. Dean, a graduate student in MIT’s Department of Economics; and Elizabeth Spelke, a professor of psychology and researcher at the Laboratory for Developmental Studies at Harvard University.

It’s a numbers game

The results bear on the question of how early-childhood educational interventions can help poor children access the same educational concepts that more privileged children have before entering primary school.

Spelke, an expert in cognitive development among children, notes that around age 5, children “transition from developing knowledge in a common-sense, spontaneous manner, to going to school, where they have to start grappling with formal subjects and building formal skills.” She adds that this can be a highly challenging transition for children living in poverty whose parents had no schooling themselves.

To address that, the researchers developed a field experiment involving 1,540 children, who were 5 years old on average and enrolled in 214 Indian preschools.

Roughly one-third of the preschool children were put in groups playing math games exposing them to concepts of number and geometry. For instance, one game the children played allowed them to estimate numbers on cards and sort the cards on that basis.

Another one-third of the preschool children played games that focused on social content, encouraging them to, for instance, estimate the intensity of emotional expressions on cards and sort the cards on that basis. In all, the games were “fun, fast-paced, and social” and “encouraged a desire to play together,” Dillon says.

Meanwhile, the final one-third of the preschoolers had no exposure to either type of game; these children formed another control group for the study.

The researchers then followed up on the abilities of children from all three groups, soon after the intervention, as well as six and 12 months later. They found that even after the first year of primary school, children who had played the math games were better at the skills that those games developed, compared to children from the other groups. The intervention using social games had effects on social skills but did not produce a comparable effect on math skills; the effects of the math games were specific to their math content.

Despite these effects, the early exposure to numerical concepts such as one-to-one correspondence, and geometrical concepts such as congruence and parallelism did not produce an advantage for the first group of students when it came to achievement in primary school. As the paper states, “Although the math games caused persistent gains in children’s non-symbolic mathematical abilities, they failed to enhance children’s readiness for learning the new symbolic content presented in primary school.”

Not adding up

The researchers have been analyzing why the intervention did not produce improvements in school results. One possibility, Duflo observes, is that children in Delhi primary schools learn math in a rote style that may not have allowed the experiment’s set of games to have an effect. Kids in these schools, she observes, “are [only] learning to sing ‘1 times 1 is 1, 1 times 2 is 2.’” For this reason, Duflo notes, the greater understanding of the concepts provided by the preschool math games might be more beneficial when aligned with a different kind of curriculum.

Or, Spelke puts it, “the negative thing that we learned” from the study is that lab work is not necessarily “sufficient to establish what actually causes knowledge to grow in the mind of a child, over timespans of years in the environments in which children live and learn.”

With that in mind, the research team is designing follow-up studies in which the games will segue more seamlessly into the curriculum being used in a particular school district.

“We want to include in the games themselves some element of bridging between the intuitive knowledge of mathematics and the formal knowledge they will be actually exposed to,” Duflo says. J-PAL is currently engaged in developing projects along these lines in both India and the U.S.

The larger goal of helping disadvantaged preschool children remains intact, Duflo emphasizes: “If we could take the poorest kids and instead of sending them to school with a [learning deficit], because they haven’t been to preschool or been to very good preschools, or their parents have not been able to help them out in the schoolwork, why couldn’t we try to use the best cognitive science available and bring them to school with a slight advantage?” 

de MIT News http://ift.tt/2stV09T