a think machine

Revamping!

2008-11-28T12:52:00.000-08:00

I am doing major revamp to this blog. New blog will focus on natural language processing / machine learning, social media research, mobile applications, semantic web, industry analysis and product management.

Some lessons from a serial tech entrepreneur

2008-08-02T19:57:00.000-07:00

Some lessons from a dinner with a serial entreprenuer.

1. Whenever there is change, there is opportunity. Always watch for big social, technological changes.

2. Adjust constantly. There will be countless times in which you will have to change and adapt your plans.

3. Be objective as much as possible. This is a very common mistake that kills entrepreneurs. Never fall in love with your own idea. You think it is good, not necessarily mean other people would think so. Try to verify it from as many as angles as possible.

4. Think about ideas that could be implemented now. Market and technology must be matured or about to mature.

5. Try not to go to markets under spotlight. Crowded market means competition. You have to be exceptionally good to surpass your peers. Finding blue ocean is the key. E.g., mobile is too hot now. You'd better avoid it.

6. Salesmanship.

7. Put a lot of efforts on hiring the first few employees. That could either make a startup prosper or kill it...

8. Network effect.

9. You need domain knowledge to compete. Don't go to market you don't know much about. If you really want to do it, hire some body who knows it.

It is not tomorrow, or the day after tomorrow. It is now..

2008-04-17T17:05:00.000-07:00

Let me make it concise. Chief Scientist of Powerset, Ron Kaplan, went to Columbia to give a talk today. I also got a chance to chat with him privately. I was impressed by their technology (I would not go into detail here). Semantic Web is not tomorrow, or the day after tomorrow, it is today, it is now...

The rise of social networking, or the so-called Web 2.0 took off at around 2003, and takes five years to mature. Though there are still a lot of things to figure out, but players are established.

Semantic Web in 2008 is like Web 2.0 in 2002 - Products are launching, people are talking and exploring. If Web 2.0's pattern continue to hold, Semantic Web will mature in four - five years. 2012, that is probably the year when everything gets set up, and shuts the door for new entrants.

China is usually two to three years later than US. E.g., Baidu was founded in 2000, while Google was in 1998, and some other search players in US were even earlier. However, the competition is different today. Internet makes information flow much much faster. Within the next two years, I believe there would be companies in China starting to work on Semantic technologies. Few people in China today study serious Natural Language Processing, which is exactly like Information Retrieval (What search technology is called in computer science) was a few years ago. But a lot more people are monitoring the internet industry.

I am excited. I am nervous. Another war in the Internet is starting... There is also gonna be a battle ground in China. And this time, it will be a lot more intense. What will happen?? What should we do??

If you are interested in the whole spectrum of "Semantic Web", check out Spivack Nova, the CEO of Radar Network's blog.

Also check out Powerset's beta.

Problem Solvers

2008-03-24T11:47:00.000-07:00

High Tech Entreprenuers are problem solvers. I have listened to most episodes in Stanford Entrepreneurial Thought Leadership program, and this definition appeared again and again. Seasoned entrepreneurs see problems as opportunities and adopt systematic strategies to tackle them. Look at most revolutionary ideas, they all attempt to solve a burning problem. Google helps to search/organize huge amount of information on the web, otherwise it is a pain to browse through the directories. RSS pushes the information to the user automatically, otherwise it is a pain to hop from one blog to another. Graphical User Interface presents the information more vividly, otherwise it is a pain to read static dry texts ... The bigger the pain, the more significant the problem, the more attractive the business opportunity.

Problems could be even seen beyond the consumer level. What are the most urgent world problems today? Poverty, energy depletion, environment, religious conflict, deadly disease ... To name a few. These problems imply vast social and business benefit. Technologists who manage to solve these problems would become the next generation of legends. Internet was an example in the past, it solves the problem of global communication and raises up a new battle ground. However, problems on the Internet landscape are becoming incrementally smaller today. But in other fields, problems are pervasive. E.g., say environment. Environment issue is hot and will become hotter tomorrow. Telsa Motors attemps to solve this problem by rolling out electronic cars. Old solution, new technology, right timing and abundant social attention, bang! Potentially a phenomenal success story.

To be a problem solver is to promote a way of thinking. Start from the problems, analyze the opportunities, decompose the problems, and tackle the pieces one by one. To be a problem solver implicitly implies to carry more responsibility and tackle the difficulties for others. It is for the consumers; it is for the local ecosystem; it is for the country; and it is even for the world - if you live in the valley. There are plenty of problems in China today, the new generation of Chinese high tech entrepreneurs maybe should inherit from this spiritual root, and assume bigger responsibilities in the money-making mission statement. From a pragmatic perspective, it is "politically right" in China of assuming responsibilities.

Elements of Sustainable Companies (From Sequoia Capital)

2008-03-21T17:29:00.000-07:00

Start-ups with these characteristics often foretells the success of a business and the likelihood of it becoming a sustainable, enduring company. We like to partner with companies that have:

Clarity of Purpose
Summarize the company’s business on the back of a business card.

Large Markets
Address existing markets poised for rapid growth or change. A market on the path to a $1B potential allows for error and time for real margins to develop.

Rich Customers
Target customers who will move fast and pay a premium for a unique offering.

Focus
Customers will only buy a simple product with a singular value proposition.

Pain Killers
Pick the one thing that is of burning importance to the customer then delight them with a compelling solution.
Think Differently
Constantly challenge conventional wisdom. Take the contrarian route. Create novel solutions. Outwit the competition.
Team DNA
A company’s DNA is set in the first 90 days. All team members are the smartest or most clever in their domain. “A” level founders attract an “A” level team.
Agility
Stealth and speed will usually help beat-out large companies.
Frugality
Focus spending on what’s critical. Spend only on the priorities and maximize profitability.
Inferno
Start with only a little money. It forces discipline and focus. A huge market with customers yearning for a product developed by great engineers requires very little firepower.

Semantic Web at Web2.0 Summit

2007-10-21T11:23:00.000-07:00

In the latest web2.0 summit, more excitement about semantic web are coming out! Besides powerset, two more companies officicially announced their products, twine and freebase. Twine is a knowledge orgnization tool, with which you can mark any interested document on the web, then the twine engine could automatically discover the semantic tags in it and merge them into the your knowledge center. Freebase is an open semantically-tagged database.

What is more exciting is that now a lot more people are paying attention to the semantic/intelligent future of web. Here are some very insightful quotes (http://www.blogger.com/post-create.g?blogID=8404935373453506700):

"Tim O'Reilly (The guy popluarlized the concept of web 2.0) brings up Google, Flickr interestingness and how users can influence results (collective intelligence etc), but that it's usually passive and hidden behind silos. But these new semantic apps are more open and they're platform players. "

"Barney (Powerset) says the real value is "making explicit what was once implicit" (in terms of data)"

"Hillis (From freebase) says Web 1.0 was a "web of documents", and that it will be the same with semantic apps - it doesn't make sense to have silos of data."

This is a graph from Nova's blog, which presents his vision of the future of web, basically pretty optimistic about semantics.

Yao Wei, Entrepreneurship and etc...

2007-10-17T15:31:00.000-07:00

A few days ago, I saw a summary of a Yao Wei's presentation in Cao Liang's blog. I know him since the business plan competition. As a veterant in entrepenurship, his dramatic style is too impressive to forget. It triggered me a lot of memories back in ITP. Now in graduate school, it is so hard to find opportunities to engage in some deep conversations like that. Basically, what I do everyday is sleep, lecture, projects and sleep again... After it is gone, I realized that how precious is an atmosphere like that, and how hard it is to find it back.

I read a lot of friends' blog today, I am so glad, or even so moved to see that the passion, comaradeship and the pursuit of bringing something different to the world is still there, although we are spreaded in all kinds of industries now. I do appreciate some people's efforts (park?spiritcell?Hong?) of building http://www.iteper.net/ . It is a great way to stay in contact. Hope that we could build it to last.

As far as what I have seen now, new york is not the kind of place that a high-tech guy would want to stay. Finance looks like the center of everything here. In the past time in investment bank, you have to command communication to earn a living. Bu these years, quantitative finance has opened the doors to a lot of Chinese guys who is just-so-so in communication but super-smart in math or programming. On these positions, although uncomparable to traders, they make big chunk of money. Say, a CS master could earn 70,000-80,000$ per year, but a quant-support (which is a programmer for real quant/trader) could easily make 100,000- 150,000$ per year and has a much steeper salary curve. This is pure temptation, as it is highly possible to transfer from CS and other engineering discplines to these professions.

A lot of people are seeking this transition. I was tempted too. But when I try to imagine what is going to happen if I choose to go into that industry 10 or 20 years later, I hesitated. I would be a middle-rank manager on a deep hierarachy, and earn probably half a million per year- so what? Do I have any impact? Can I bring any positive changes to the world? Who do I serve besides my boss? It looks that this is not the kind of life I want.

However, I believe there would be friends who want to get into this industry. NYC is definitely the place you should come. A lot of my friends here in finance are living on the edge. They understand the brutality of job market, they are stressed out, they are building connections and they are competing to stand out. This is probably the right attitude and nerve that some one needs to survive in finance. So come to arena.

A weak intelligence era

2007-10-17T11:07:00.000-07:00

I believe a new fad in the IT industry is emerging. I would call it as a weak intelligence era. The weak/strong distinction comes from an old argument in the AI field. Weak here means to reach the result of behaving intelligently without imitating the process of thinking "intelligently".

This fad is mainly based on latested advances in artificial intelligence and the huge amoung of data generated after web2.0. A long-time controversal vision - the semantic web - may also help to shape this new era.

From a consumer's view, applications with weak AI will:
1) Know who you are, what you like, what groups you belong...
2) Be able to predict your preferences.
3) Engage in simple human-computer interactions (Verbal/NLP, visual/Computer Vision...).
4) Possess a sense of autonomy.
5) Build a weak level of personal relationship.

From a business's view, applications with weak AI will:
1) Make more sense of various kinds of data.

When will the fad come? I think the question lies in when there could be several killer applications that dramatically show the improvement of the cluster of new methods. It will not be long~

Recommendation Systems

2007-10-13T20:16:00.001-07:00

Check out www.mspoke.com, these guys are doing recommendation systems based upon some machine learning techniques. Some latest acqusitions as eBay's acquisition of StumbleUpon , CBS acquisition of Last.fm and this weekend's MSNBC acquisition of Newsvine seemingly indicate that recommendation system will be a big thing (From Read/Write Web).

I have personally tried mspoke's product - www.feedhub.com which is a filter to generate personalized RSS feed, but I did not find it particularly useful. A user like me reads feeds for fun, so I do not care how many unread articles are out there. Its target should be some more serious readers who have very tight schedule. However, would these group of users trust a machine to filter information for them? Hopefully, AI would advance to the point of obtaining that trust.