Everything should be made as simple as possible, but no simpler -- A. Einstein
Jing Conan Wang is a Principal Scientist and leads the machine learning modeling of Intellimize, whose AI dynamically personalizes mobile and desktop websites for each unique visitor in real time – moving prospects quickly through customers’ funnel and increasing conversion rates.
He was a Senior Engineer at Google Brain and worked on improving recommender systems using reinforcement learning and deep learning between 2018 and 2019. Prior to Google Brain, he worked at Google Ads between 2014 and 2017, where he drove 20+ modeling launches that bring Google more than $200M revenue annually. He obtained Ph.D. degree from Boston University (advisor: Yannis Paschalidis) and B.E degree from Huazhong Univ. of Sci. and Tech.
He has good knowledge of mathematical modeling. He received prize of Honorable Mention in Mathematical Contest in Modeling (MCM) 2009, and second prize in Chinese Undergraduate Mathematical Contest in Modeling (CUMCM) 2009. He has extensive experience of research and open source development. He was a participant of Google Summer of Code 2012 and a mentor for Google Summer of Code 2013, both for the Honeynet project. He was a finalist of Microsoft Young Fellow in 2009.
Data infrastructure serves two purposes at a high level: 1) to help business leaders make better decisions through the use of data (analytic use cases) and 2) to build data intelligence into customer-facing applications, including via machine learning (operational use cases).
Two parallel ecosystems have grown up around these broad use cases. The essay mentioned there hasn’t been consensus on the two ecosystems will converge eventually.
My take for that is that most of the two ecosystems’ infrastructure will converge because there is no inherent boundary between the two use cases. For example, interactive data analysis is essential to build new ML features or inspect models’ quality. Building interpretable models are also super helpful to get business insights.
Instagram is a phenomenal product of the mobile Internet. One interesting thing from this interview is how tools and technologies made Instagram possible. In particular, two key factors enabled Instagram. 1) The popularity of mobile phones with good cameras, and 2) the availability of open-source tools and scalable AWS cloud infrastructure. The former created the mobile phone-sharing market of billions of users, and the latter made it possible for Instagram to scale to serve a massive amount of users with a small team.
Both Kevin and Mike are very excited about what is happening in the data & machine learning field. Many data infrastructures are being made accessible for the general audience, and the cost of setting up a machine learning (ML) system has reduced significantly in the past years. Open-source tools like TensorFlow and many clouds AI infrastructures affect the tech community, similar to what the AWS had influenced mobile development. As the machine learning infrastructure matures, what matters more for a company is what problem you use ML to solve and how you tailor your system to solve those problems.
Roadmap: Consumer Earthquakes. In this video series, BVP Partner Kent Bennett details what makes a consumer earthquake startup and the keys to a viable business model and long-term defensibility.
An innovative product organization’s core building blocks should be “pizza-teams,” which are small teams that are often 5-8 people and formed by people of different functional groups. Pizza teams could be either permanent or temporary depends on the nature of the problems that need to be solved.
Each pizza team needs to be fully aligned with the organization’s direction. However, day-to-day decisions should be made autonomous instead of flowing through the management reporting chain, which slows down the feedback loops and often yields suboptimal results.
Functional managers should engage in coaching and empowering instead of telling their reports what to do. Managers must step back to create this space while stepping in to remove impediments, clarify context, and provide guidance.
A good framework for managing your time is important for personal productivity. One interesting framework is the Tripartite Productivity System. The key is to handle three challenges: managing information, managing tasks, and managing time.
Information: We need a place to retrieve information, to process them, and to store the distilled knowledge. Generally, any document processing or note-taking software (i.e., “Filing Cabinets“) should work.
Task: This is to track the context of each task to allow attention to be focused on taking action instead of recalling contexts. There are plenty of tools for this. I personally use Trello as a personal Kanban system.
Time: We should align our day-to-day tasks to a larger picture by tracking and reviewing our time usage.
Reviewing: It is also important to do a weekly review during which you could reflect on what you have done well and what could be improved in the past week.
How to debug remote work, as suggested by new research. With the pandemic, a lot of startups have to operate in the full-remote mode for an extended period of time. This great article by Atlassian summarizes the ways that could improve the productivity of WFH teams. Here are several highlights from the article:
We have an opportunity to embrace the fact that hours worked (or cold-calls made, or lines of code written) was never a meaningful performance metric in the first place. Now is an ideal moment to shift our attention away from outputs of effort and focus on outcomes achieved instead.
Employee motivation is much more critical in the WFH setting. You have to make sure employees have intrinsic motivations in order for them to keep staying focused and delivering results. This is also where innovation would come.
Handle the inequality caused by working from home. Different people have different expectations of working from home. Adjust accordingly. 1) Household complexity. 2) Role complexity 3) Network quality.
WFH removes the isolation between work and life and causes a lot of extra challenges. Be mindful of the different situations employee need to face at home and make sure it doesn’t create systematic inequality.
In this blog, the former CircleUp CEO Ryan shared his journey of creating the CircleUp and his decision to step down as CEO of the company. This is a super honest reflection with a lot of details. A good read for anyone who is curious about the startup world.
The current COVID-19 pandemic is not the first pandemic, and it won’t be the last one. The book The Great Influenza by John M Barry may help you envision how the future years or even decades look like. I learned this book from Bill Gates’s article in May but only finished it recently. Despite that a whole century has passed, a lot of things in this pandemic have happened in the same way as the 1918 one so the book is still very relevant.
The book is not just a chronicle, it illustrates how American society reacts to a challenge like this and provides many interesting observations. In Chapter thirty-two, the author speculates that it was influenza that caused the bad judgments of the US President Woodrow Wilson in the Paris Peace Conference and caused the even deadlier World War II. In early 1919, the representatives of all nations gathered in Paris to discuss the new world order after World War I ended in 1918 partially due to the pandemic itself. For weeks and then months, heads of the US, Britain, and France were negotiating on the terms, and the sessions often went brutal. On April 3, Woodrow Wilson suffered a health attack, which Barry thinks was influenza, and had a fever of over 103 degrees. Partially because of the deteriorated health, Wilson lost the grit and ceded the extremely harsh terms insisted by the French prime minister Clemenceau, which created chaos in Germany and caused the rise of Hitler. Wilson also agreed to Japan’s insistence that it takes over German concession in China, which triggered the May Fourth Movement in China and catalyzed the spread of communism throughout the country.
It is incredible that the aftermath of the 1918 pandemic could be felt for the whole century. Had the Wilson not cede in the Paris Peace Conference, the rest of the 20th century will be quite different. Now that the current President and many white house officials have also contracted COVID-19 recently and an election is on the horizon, there is a chance that the aftermath of this pandemic may be equally long.
One thing I find interesting is how Daniel’s perspective on time management and meetings. According to Daniel, a meeting would be a waste of time if one or more of the elements are missing. The make-up meetings are the single largest source of optimization for a company.
Another interesting thing from the interview is his comments on learning. Learning is all about the abstraction of the world. As you keep trying things, you will figure out what’s important and what’s not and build an abstract model yourself.
There are many tools to help with the process. One thing I find very useful is to organize knowledge in a graph and materialize the graph in some places (i.e. build your personal wiki). Previously I have been using Google docs to manage my notes and add cross-reference among them. Recently I started to use Notion, which I highly recommend. This video from Notion is a good tutorial for building a personal wiki using Notion.
The following tweet from Joaquim shows the first snowball fight in 1896 that happened in France. This is probably the most well-dressed snowball fight I’ve ever seen.
This cool tweet from Simon shows the trajectory of all active satellites. One interesting thing is that you could see the trains of StarLink satellites very clearly in the animation.
This week I recommend the book Behind the Cloudby the Salesforce CEO Marc Benioff. The book describes the history of Salesforce, a legendary Software-as-a-Service (SaaS) company, and summarizes Marc’s lessons on his journey.
Before Marc co-founded Salesforce, he had worked on enterprise software for many years as an Oracle executive. During the Internet boom, he took an extended leave from Oracle, spending most of the time in Hawaii and then India, to reflect his career and to think about his next venture. He came back with a belief that that software could be delivered through the Internet instead of manually installed by customers. He created Salesforce to fulfill this belief.
March Benioff used a very creative marketing strategy in the early days of Salesforce. Instead of defining Salesforce as another enterprise software company, Marc marketed Salesforce.com as a new type of company that is fundamentally different from its predecessors — a Software-as-a-Service (SaaS) company. He waved the “No Software” marketing campaign and claimed enterprise software was dead. He also organized protests during the events of his software competitors. The controversial move drew a lot of eyeballs.
His strategy was a big success. The Software-as-a-Service has become a huge market, and its market size is projected to be 307.3 Billion by 2026. There have been a lot of SaaS IPOs in 2020 so far, including Snowflake, JFrog, Sumo Logic (Please see this Crunchbase report for the full list). As the first and still largest SaaS company, Salesforce is one of the examples for companies that define new categories of markets.
The opinions of the two Marcs, despite their wordings, are referring to the same trend. Marc Andreessen is talking that software will be more and more important. Marc Benioff is talking about how the delivery of software will shift to the cloud.
Lindy Effect refers to the fact the life of a thing is likely to be proportional to how long it has already existed. Intuitively, a thing that exists x years is likely to exist for another x years. This is the reason why reading news is not that useful and we should focus on digesting information that is more time-proven.
I highly recommend the Netflix hit The Social Dilemma and here is a good review from the New York Times. This documentary from Jeff Orlowski explores how addiction and privacy breaches are features, not bugs, of a lot of Internet products. Although the title calls out social networks, this problem exists for many other Internet products built on advertising business models, which sell users’ attention to advertisers.
The main benefit of advertising is that it allows free access to information. Without the income from advertising, creators would have to hide their content behind paywalls, which makes them less accessible to lower-income people and widens the information inequality in society. However, advertising also misaligns the interests of the companies that build products and their users. Without enough counter-balance, companies will inevitably put the interests of their real customers — advertisers — before their users.
One important counter-balance factor is competition — companies that put less emphasis on users will lose market share, which reduces their values to advertisers. However, when the market share is concentrated on the top players, companies will tend to optimize for short-term interests and cause problems both for their users and the whole industry.
This is not the first time we face such a situation. Newspaper, Radio, and TV Broadcasting are all built on the same premise as the Internet. History tells us that the public and the government need to play a more active role in regulating the market when it becomes mature. If you are interested in this topic, I highly recommend you read the book The Attention Merchants: The Epic Scramble to Get Inside Our Heads by Tim Wu.
One difference between the Internet and its predecessors is that it has unprecedented power to influence users thanks to the big data and advance of machine learning. TikTok is an example of this and this podcast by Eugene Wei and Sonal Chokshi covers Tiktok from algorithmic and product design perspectives.
Let us switch the topic to the platform that created the hit. I highly recommend you to read this book to No Rules Rules: Netflix and the Culture of Reinvention. Netflix has been famous for its innovative working environment. In the past two decades, it has done two successful transformations — first from DVD rentals to streaming, and then to original contents. In this book, the co-founder of Netflix tries to decode the culture genes of Netflix and how he made Nextflix an innovation hub by increasing talent density and removing processes relentlessly. Here is the podcast of Reed Hasting in a16z about the book. Another interesting content is the dialog between Reed and Chris Anderson.
Reed is not the only founder of Netflix. This video is by another co-founder Marc Randolph about the early history of Netflix.
Tabby’s Star. In 2016, astronomer Tabetha Boyajian announced the discovery of weird behavior coming. The star exhibited odd dips in light, which are very irregular and unpredictable. As scientists were not able to figure out the cause, some speculate it is a Dyson sphere created by an alien civilization. Although there are some new pieces of evidence suggesting this is unlikely, people couldn’t stop speculating before the science community could fully understand what happens.
The “Notorious RBG” Ruth Bader Ginsburg, the second woman who served in the Supreme Court and a pioneering advocate of women rights, has passed away this week. Her pass-away is a very big loss for both her families and the whole country.
Although the Supreme Court is supposed to be insulated from politics, the replacement of her position likely will stir another round of political drama in the country. This great article from the Guardian explains what does Ruth Bader Ginsburg’s death means for the supreme court. The US supreme justice nomination is critical for both parties because supreme court justices have life-terms thus could influence policies over a long period. Trump has already made two nominations in his first term.
Given there is only one-and-half month to the upcoming election, Republicans have no intention to wait. In his mourning statement to RBG, the majority leader McConnell has stated that President Trump’s nominee will receive a vote in the Senate. Ironically, during the case of Merrick Garland Supreme Court nomination in Obama’s last year in the office, the republican refused to consider Obama’s nomination citing it was an election year despite that there were still eight months to the election. However, the gamble move will likely provocate more liberal voters and may swing some conservative voters in the upcoming election.
TikTok and WeChat Ban
The drama of TikTok continued this week. TikTok rejected Microsoft’s bid and partnered with Oracle. The details of the TikTok-Oracle “deal” change by the day, the latest update is that Oracle would only be a partner to provide technical support. At the same time, Bytedance will keep the majority stake but will appoint an all-American board and seek a public listing in the US. At the same time, the department of commerce issued the WeChat, TikTok US shutdown order. Users won’t be able to download both apps from Sep. 20. Wechat won’t get “any provision of internet hosting services enabling the functioning or optimization” immediately from Sep 20. TikTok, however, will be treated differently and its deadline for service shutdown was thoughtfully delayed to be after the election, indicating the current administration has put the problem under the re-election calculus.
Contrary to what most people think, the owner of Instagram–the largest competitor of TikTok–ardently opposed the ban. His point is valid. The US companies have thrived because of the borderless nature of the Internet that past US administrations have supported. In contrast, the Chinese government ardently supports the “Internet sovereignty” concept and wishes to control information flow in-and-out of its borders. The ban has put the US and China in the same camp, and more and more countries may follow the lead and two countries. When that happens, it will be the US rather than Chinese companies to take a toll.
Bessemer Venture Partners has released its investment memos for its top investments, including big names like Shopify, Pinterest, etc. The well-written memos contain a lot of useful information about the BVP’s investments and the early history of the invested companies. I highly recommend every further entrepreneur or investor to read those memos.
Zanclean flood – Wikipedia. The Mediterranean Sea has been the cradle of civilization. It was the remains of an ancient sea but dried up after being cut from Atlantic Sear due to the high evaporation in the area. Scientists believed that at about 5.33 million years ago, the strait of Gibraltar was open and created a river whose volume is about a thousand times larger than the discharge of the Amazon River. The river filled the sea in a few months to 2 years of period. It is fascinating to see that an area of 1 million square miles could be filled in such a short amount of time.
Another interesting thing about TikTok is that it’s the first social network platform to reach popularity both in Chinese and International Markets. This is very unique because previous generations of successful Chinese Internet companies copied their business models from the US, and their successes were mostly limited to China.
TikTok is also much more than a social app. For this topic, I recommend you to read The Video-First Future of eCommerce, a great essay by Connie Chan and Avery Segal on the video eCommerce popularized by TikTok.
Amazon and the impact of technology on gig workers
Let us switch the topic to the largest eCommerce player in the US. Amazon’s Profits, AWS, and advertising is a great essay by Benedict Evans about Amazon’s business models. One interesting fact is that Amazon is much more than just an eCommerce platform now (a pattern for big companies!). In fact, in addition to eCommerce, it becomes the largest 3rd party eCommerce platform and the largest cloud provider, and it also has a thriving “search Ads” business that threatens Google.
As the big tech companies become more and more powerful, they become more and more impactful to human life — not only for their employees/customers but also equally important for their gig-workers. One interesting article I saw this week is that Amazon Drivers Are Hanging Smartphones in Trees to Get More Work. This may sound funny to us, but the people who are doing this are very serious because they depend on the platform for their main income.
When the rules of the platform are unfair to its gig workers, it also will significantly harm their life qualities and eventually backfire the platform. This week a viral article about the miserable life of the couriers who work for Meituan-Dianping — the uber eats in China — puts it under the fire. Please see this TechCrunch report for more details.
Maya Lin is the FIRST woman to design a memorial on the National Mall. In 1981, Lin, then still was an undergraduate, submitted a proposal for the Veterans Memorial, to be built on the National Mall in Washington D.C. H. The design was initially controversial for several reasons: its untraditional design,, Lin’s Asian ethnicity, and her lack of professional experience. Lin defended her design before the US Congress, and a compromise was reached: The Three Soldiers, a bronze depiction of a group of soldiers; and an American flag were placed to the side of Lin’s design.
Success! You're on the list.
Whoops! There was an error and we couldn't process your subscription. Please reload the page and try again.
The Global AI Talent Tracker. This great analysis from macropolo.org shows how AI talents migrate among different countries. The United States has a large lead over all other countries in top-tier AI research, with nearly 60% of top-tier researchers working for American universities and companies. The result also clearly shows the US lead is built on attracting international talents as more than ⅔ of the top-tier AI researchers received undergraduates in other countries.
AI development often feels “closer to molecule discovery in pharma” than software engineering. This is because AI development is a process of experimenting and the job of an AI developer is to fit a statistical model to a dataset, test how well the model performs on new data, and repeat.
One of the complexities in the real-world data is that a lot of them have long-tail distribution. Intuitively, what this means is that that there are a lot of edge cases that appear infrequently yet important. This essay presents a guideline to tackle the issue and other similar ones.
Another interesting thought that I have after reading this essay is that the common metaphor of “data is oil” is not quite accurate. The energy of one barrel of oil is always the same regardless of the amount of oil you have used. In contrast, the benefits of extra data diminish as the size of data increases. The reason is that although training with more data helps the handling of new edge cases, the chance of seeing new cases decreases as you move further down to the tail. As a result, it is an art to find a balance of the predictive power and the complexity when we decide how much data we need.
AI job listings plummet as COVID-19 recession appears imminent. François Chollet, the creator of Keras and a researcher at Google Brain, posted a tweet that shows that AI jobs plummeted since the start of the COVID-19 recession. One of the reasons is for the correction of the hype in the past few years. However, this doesn’t mean that we should be bearish for the future of AI. Instead, COVID19 has been a catalyst for technology adoption for many industries (e.g., Zoom for video conferencing). I am optimistic that the pandemic would actually make more traditional companies start to embrace AI technologies.
📌 We (Intellimize) are hiring for ML & AI scientists/engineers. Please shoot me an email if you are interested.
A lot of teams have moved to be fully remote during the pandemic. This blog lists a lot of virtual games that may be helpful to increase the bonding of the team and. In addition to those listed in the blog, I also highly recommend the drawphone game, I have tried it many times and enjoy it a lot.
🍭 Interesting Facts:
The Salton Sea is the largest lake within California and it is fascinating to see that it was created only in 1905 purely because of an engineering mistake. Unfortunately, it is dying because of the reduction of the input water recently.
I hope you’ve had a great week! Here are the recommendations for this week:
Making sense of MVP (Minimum Viable Product) – and why I prefer Earliest Testable/Usable/Lovable. This great essay from Henrik Kniberg explains minimum viable product (MVP), an important concept in lean development. An MVP is a product that is released to collect customers’ feedback as early as possible for product iteration. Because the main purpose of an MVP is to learn customers’ feedback rather than to build the final product, it needs to be minimum yet COMPLETE. The figure above is a metaphor for the idea: if your goal is to build a sedan, your MVP should be a skateboard, which helps collect customer’s feedback on transportation tools, instead of a perfect tire. From the early adopters who try the skateboard, you may learn that customers want to enjoy fresh air while moving. As a result, you could build a convertible car that fits customers’ needs better than the original design.
This essay from Benedict Evans gives concrete data about the surge of e-commerce during the pandemic in the UK and the US — The UK e-commerce penetration went from 20% to over 30% in two months, and the US from 17% to 22%. Prior to the pandemic, e-commerce penetration in the two counties has been increasing at a quite steady pace. However, the pandemic lockdown has pulled forward a huge amount of future adoption within a short period. It would be interesting to see how much of those adoptions would stick to.
As the e-commerce boom continues, Amazon itself has become one of the largest shipping carriers in the world and shipped 415 million packages in July alone. Besides, other US shipping carriers also saw a surge in the volume. UPS saw volume grow 26% in July compared with the average monthly growth of 23% in the April to June period. FedEx volume rose 22% compared with 19% average growth in the first three full months of the coronavirus pandemic.
Many software companies are lining up for IPOs as the capital market is booming. One particular interesting thing is that the percentage of enterprise software companies is super high in this round of IPOs. In the past decades, the enterprise and consumer software industries in the US have diverged into two directions. Most of the consumer businesses like Facebook and Google build walled gardens, making it very difficult for new players to enter the market. In contrast, enterprise software companies are embracing specialization and eagerly integrating with other enterprise vendors. This makes it much easier for new enterprise startups to reach IPO stages. Besides, this review of high-growth SaaS IPOs in 2019 would give you an idea of the enterprise companies that went public in 2019.
Highly recommend this video created by Phil Libin, the co-founder of Evernote and recently the mmhmm app, to explain the motivation behind the new app’s name. The mmhmm app has been generating a lot of buzzes recently but its name sounds ridiculous to many people. In this video, Phil describes his philosophy of the naming: A ridiculous yet interesting name is way better than a boring one. It is better to create a product that is loved by someone and hated by some others than a product that nobody cares about.Besides, mmhmm is really fun to use. If you haven’t tried it out yet, you should!
The Passion Economy and the Future of Work. We are in a world that the definition of jobs has changed significantly, and the process has been accelerating since the pandemic. In the new world, individuality is respected and encouraged, and people could and should earn their living from their individuality.