Author | Bao Yungang (Researcher at the Institute of Computing Technology, Chinese Academy of Sciences)
Recently, the banning of the engineering tool Matlab at Harbin Institute of Technology and Harbin Engineering University has sparked ongoing discussions.
On June 19, Bao Yungang, a researcher at the Institute of Computing Technology, Chinese Academy of Sciences, shared his thoughts on this incident during the “CCF YOCSEF Hangzhou: Special Forum on the Ban of Matlab” on Weibo.
Bao pointed out that while we are thinking about how to address the urgent issue of “Matlab being banned,” we need to consider how to create tools like Matlab in the future—technologies that can dominate the market.
By reviewing the 40-year development history of Matlab, Bao believes we must change our mindset in four key areas.
The four concepts include: creating practical tools instead of just aiming to publish papers; utilizing the tools instead of discarding them after completion; applying teaching scenarios rather than viewing teaching as a burden; and establishing a mindset for long-term commitment instead of expecting quick wins.
The following is the full text of Bao Yungang’s speech at the “CCF YOCSEF Hangzhou: Special Forum on the Ban of Matlab,” reprinted with permission from ScienceNet.
Many people are pondering how to solve the urgent issue of “Matlab being banned.”
They have reflected on the painful development of industrial software in China, criticizing domestic issues such as piracy, intellectual property protection, and the tendency to emphasize hardware over software. Many good suggestions have been made, such as providing open-source software alternatives for various Matlab functions.
These are all matters of “yesterday and today.”
Now I would like to discuss matters regarding “tomorrow.”
Each of us can ask ourselves a question: Starting from this moment, given 10 years or even 20 years, can we create something that dominates the market? (Not implying we should literally dominate others, but rather create something indispensable for others.)
If viewed from this perspective, the incident of “Matlab being banned” may inspire us more— we all know that Matlab was initially just a small tool used for teaching by Professor Clever Moler at the University of New Mexico in the 1970s. So why has it become a crucial tool decades later?
Let’s outline several concepts reflected in the development of Matlab: 1. Create practical tools instead of just aiming to publish papers.
On the Matlab website, there is a brief history of Matlab written by Professor Moler in 2018.
He stated that in 1971 and 1975, his team applied to the NSF for two projects aimed at “exploring the methods, costs, and resources for developing high-quality mathematical software.”
He himself believed that to some extent, these two projects were failures because they did not publish a paper; they only developed two software:
one was EISPACK, and the other was LINPACK.
Moreover, these two software cannot be considered particularly innovative academically, as EISPACK merely translated algorithms written in Algo60 from papers published between 1965 and 1970 into Fortran, while LINPACK simply rewrote them in Fortran.
2. Utilize tools instead of discarding them after completion.
Although EISPACK and LINPACK lacked papers and seemed to have low academic innovation, they were indeed two very useful software.
The EISPACK development team wrote a user manual in the 1970s, which has been cited over 1800 times on Google Scholar and was widely used in the 1970s and 1980s.
LINPACK is even used as the benchmark testing program for the Top500 list of supercomputers, significantly influencing the development of supercomputers worldwide.
3. Apply teaching scenarios instead of viewing teaching as a burden.
Matlab is a product of Professor Moler’s desire to apply EISPACK and LINPACK in the teaching process.
If Professor Moler had not been dedicated to teaching, aiming to help students better grasp linear algebra and numerical analysis while easily using EISPACK and LINPACK, he would not have been motivated to create a small Matlab tool to encapsulate the interfaces of these two software for easier student use.
Today, due to the harsh competitive environment and assessment pressures in research, many view teaching as a burden, believing it detracts from research.
However, teaching is actually the best application scenario for testing new technologies and tools, as the cost of trial and error is low, and students’ creativity and initiative can help improve and optimize technologies and tools.
The commercialization of Matlab ultimately occurred because two students at Stanford University showed great interest in Matlab during Professor Moler’s class and actively proposed to rewrite it in C and port it to IBM PC.
Many technologies originally developed in classrooms, such as the RISC architecture, emerged from Professor David Patterson’s experiments in Berkeley.
4. Establish a long-term commitment mindset instead of expecting quick wins.
Sticking to one thing can yield astonishing cumulative effects over decades.
The technologies currently dominating the market in China are almost all products of over 20 years of accumulation by others.
From the first version of Matlab to now, it has been 40 years; the first generation of EDA software from the early 1980s has also been nearly 40 years; the first generation of Intel microprocessors was introduced around 1970, which has been 50 years.
Taiwan Semiconductor Manufacturing Company (TSMC) was established in 1987 and has accumulated over 30 years.
In academia, many influential works are also products of years of accumulation. For instance, the ACM System Software Award-winning software has typically been developed over decades, such as LLVM optimized for 17 years, Eclipse for 19 years, Wireshark for 22 years, Coq for 31 years, and GCC for 33 years.
Upon closer examination of Matlab and its company MathWorks, it can be said to be a model of long-term commitment.
MathWorks was founded in 1984 with only 1 employee.
The first income was in 1985 from selling 10 Matlab licenses to MIT, generating $500.
In its early years, MathWorks was rather inconspicuous, with a joke stating that the number of employees doubled each year for the first 7 years: 1 employee in 1984, 2 in 1985, 4 in 1986, until 1991, 7 years later, there were only 128 employees.
Compared to many startups today, this growth rate seems like that of a snail.
However, they focused on enhancing Matlab’s functionality, making it an industry-leading software tool.
By 1997, MathWorks’ revenue reached $50 million with 380 employees.
As of 2019, MathWorks’ revenue was $1 billion, with over 3,000 employees and more than 4 million users worldwide.
Although the revenue may not seem large, we should learn from this model of continuous accumulation.
Striving for excellence in a technology can lead to becoming an invisible champion in a specific niche.
In conclusion, while we think about how to resolve the urgent issue of “Matlab being banned,” we need to consider how to create work like Matlab in the future—technologies that can dominate the market.
This requires changes in our mindset and actions.
As for the specific changes needed, the four points I outlined earlier can serve as a starting point:
(1) Create practical tools instead of just aiming to publish papers.
(2) Utilize tools instead of discarding them after completion.
(3) Apply teaching scenarios instead of viewing teaching as a burden.
(4) Establish a long-term commitment mindset instead of expecting quick wins.
Cooperation Inquiries:[email protected]
Submission Inquiries:[email protected]
