[ad_1]
It’s no secret that algorithms run the world, powering everything from Google search results to Uber’s car-sharing capabilities. But under the hood of the world’s computing infrastructure is a more fundamental set of algorithms that have not been previously analyzed with respect to where they were created.
A new study led by MIT reveals that many of these pieces have been made in the United States, some by Native Americans but increasingly also by immigrants who work in their institutions.
Analyzing the improvements over 70 years in the 128 most important “families” of algorithms, the researchers found that approximately two-thirds of the improvements came from researchers at North American institutions, but that in the last 30 years more than three-quarters of the Contributions come from scientists from other countries.
“If we want the United States to continue to be ground zero for computer science, we must ensure that our policies facilitate the continued participation of international host researchers in our institutions.” says Thompson, a research scientist at MIT’s Computer Science and Artificial Intelligence Laboratory (CSAIL) and the Sloan School of Management.
The study shows that algorithmic progress to date has been disproportionately focused on the West. Even though North American and European residents make up only 15 percent of the world’s population, they have contributed more than three-quarters of the algorithms. The determining factor appears to be the wealth of a country: GDP was found to be more important in producing important algorithms than the actual size of a country’s population.
(On average, a $ 10,000 increase in GDP per capita produced a greater increase in algorithm contributions than a population increase of 100 million people.)
“There is a danger that algorithm development could suffer from the problem of ‘lost Einsteins’, where those with natural talent in underdeveloped countries cannot reach their full potential due to a lack of opportunities.” Thompson says.
Another key finding highlighted the importance of federal funding for university research: 82 percent of influential algorithms came from the work of nonprofits and public institutions such as universities, as opposed to private companies.
“Generally speaking, giving money to public institutions means that it is more likely to obtain a public benefit”, says Thompson, who co-wrote the study with Georgia Tech visiting student research assistant Yash Sherry and former CSAIL researcher Shuning Ge.
To develop the dataset, the team first turned in more than 1,000 research articles and 50 textbooks to create a list of approximately 300 algorithms that were the first to solve particular problems or state-of-the-art improved methods. These included everything from better listing rankings to the infamous “peddler” problem, where the goal is to find the fastest route through multiple cities.
Using the team’s 300 algorithms, they finally analyzed a subset of 180 that could be obtained to obtain information on authors and institutions. Collectively, the researchers refer to this set of fundamental algorithms as “The algorithmic commons” – because, like Digital Commons, it represents advances in knowledge whose benefits, they believe, can be widely shared.
“The best thing about the algorithm improvement is that you get more results without having to invest more resources.” Thompson says. “Just as a productivity improvement for a company allows them to produce more results for a given set of inputs, an algorithmic improvement allows a computer to tackle bigger and more difficult problems with the same computational budget.”
The project coincides with ongoing work by Thompson and Sherry showing that improvements in algorithms have often rivaled and even surpassed decades-long improvements in computing hardware that come from Moore’s Law.
Journal reference
- Thompson NC, Ge S, Sherry YM. Building the common algorithms: Who discovered the algorithms that underpin computing in the modern business? Global Strategy Journal. 2020; 1-17. DOI: 10.1002 / gsj.1393
