Support this blog.
Buy from Amazon
2008-11-20
Launched!
After months of silence, he's back.
Yes, I mean me. I'm back.
It's impossible to find time writing this blog being this close to launching such a major product with such a huge impact. If you are obsessed with Google stuff as much as you are obsessed with me, you would already know about SearchWiki.
What's the best part developing SearchWiki? Definitely on reading people's comments. So go ahead, write some comments for me to read!
Labels: work
2008-07-11
I'm Joining Google
For those of you who don't have insider knowledge (shame on you!), I have accepted an offer from Google and will be joining them shortly in their Mountain View headquarter. So yes, I will be relocating to Bay Area. And this allows me to escape from the horrible traffic in L.A. (If you want to argue the traffic is also bad in Bay Area, you have no idea what bad traffic is.)
It wasn't an easy decision. I lost my title of "Principal." (Even my other offer at least gave me a "Senior" title, although it is not a good time to join a company that has been messed around by that convicted monopoly.) I took a pay cut, made worse by knowing that I was about to get another raise. I will move from a small, noisy cubicle to sharing a cubicle with others. But the temptation of moving up north is too great to pass.
I need to thank Symantec, for an opportunity to an above average interesting job. Also for offering me to move to their Mountain View office once they knew my intention to go north. And the best of all, still paying for my CEAS registration. Free banquet!
It wasn't an easy decision. I lost my title of "Principal." (Even my other offer at least gave me a "Senior" title, although it is not a good time to join a company that has been messed around by that convicted monopoly.) I took a pay cut, made worse by knowing that I was about to get another raise. I will move from a small, noisy cubicle to sharing a cubicle with others. But the temptation of moving up north is too great to pass.
I need to thank Symantec, for an opportunity to an above average interesting job. Also for offering me to move to their Mountain View office once they knew my intention to go north. And the best of all, still paying for my CEAS registration. Free banquet!
Labels: work
2008-05-07
9-5
9-5
9-5
9-5
And I'm not talking about office hour.
Today was the longest meeting in my life. A nonstop meeting from 9am to 5pm. The meeting continued even during lunch. And, of course, it was also the most unproductive day in my life...
9-5
9-5
And I'm not talking about office hour.
Today was the longest meeting in my life. A nonstop meeting from 9am to 5pm. The meeting continued even during lunch. And, of course, it was also the most unproductive day in my life...
Labels: work
2008-05-04
Exciting Day is (Hopefully) Coming
I have finally received my bonus for being evil. The amount was 40% less than what I have expected, since I thought the amount was after tax but it turned out to be before tax. Anyway, now this thing is finally out of the way, I can start doing what I have been planning for a while. I'll keep you guys posted here as soon as I can.
Labels: work
2008-04-27
Going Cheap
Going cheap doesn't necessarily save you money. It just shows people you are cheap.
My company has one (new) microwave in the kitchen. Due to popular demand, they've added another new one. But after they are used for two months, one of them is already broken.
How could that be? Well, they probably just buy the cheapest microwaves available in the market, which turns out to be Haier. But at this rate, they are going to pay much more than spending more on a better microwave.
My company has one (new) microwave in the kitchen. Due to popular demand, they've added another new one. But after they are used for two months, one of them is already broken.
How could that be? Well, they probably just buy the cheapest microwaves available in the market, which turns out to be Haier. But at this rate, they are going to pay much more than spending more on a better microwave.
Labels: work
2008-03-21
Job Market
Even though I haven't been actively looking for new opportunities, I have been contacted directly by recruiters of two large companies to see if I'm interested in joining their advanced techonolgy groups. And I thought this is the start of a recession, where the job market is supposed to be shrinking.
In case you're interested, of the two companies I'm talking about, one has made an unsolicited bid for the other...
In case you're interested, of the two companies I'm talking about, one has made an unsolicited bid for the other...
Labels: work
2008-02-22
Free In-n-out
This week we have a free In-n-Out lunch on our company site, this time for the celebration of Black History Month. For those of you living outside the southwest, In-n-Out is a very popular and inexpensive fast food restaurant here, and their burgers are really great for their price.
This is actually the second time in a while. Different from last time, however, is that this time we need to first get a ticket to claim the meal on the trailer. Why all the trouble? That's because last time they ran out of food at around 12:30, while they claimed to last until 1:30, and someone I know got at least 7 burgers. Talking about selfishness!
This is actually the second time in a while. Different from last time, however, is that this time we need to first get a ticket to claim the meal on the trailer. Why all the trouble? That's because last time they ran out of food at around 12:30, while they claimed to last until 1:30, and someone I know got at least 7 burgers. Talking about selfishness!
2008-02-18
Hardest Part
It's difficult to realize which part of my job is the hardest until near the end of a full project cycle of my own.
It would seem hard to know what to do in the beginning. People tend not to know what they want. And it is a researcher's responsibility to understand their product, their business, and what is technically feasible for them to excel in their market.
Then it is equally hard to make them agree with your decision. Convince them that even though you don't really know what they do, you know what is the best for them, and you know what it takes for them to be successful.
What follows is the easy part, which is the actual research work. Abstracting problems. Reading papers. Designing algorithms. Implementing prototypes. Running experiments. Writing patents. Things people actually expect researchers to do.
The next step would be the boring part. And this includes periodically presenting results and getting feedback. With the only goal of making them take over what you have done. Usually it would mean horrible work like cleaning up code and, god forbid, documentations.
And just before you thought you have arrived to the fun part, namely, writing papers, the worst has yet to come. It is not about writing a high-quality paper. After all, these are industrial research projects with real-world impact that some academic conferences are dying for.
Before disclosing company's trade secret and product's internal details, you need to know if you can actually do that, and if so, how much. In the last few days, in order to decide the position of the paper, I have talked to their engineers, scientist, technical leader, product manager, PR, marketing people, and attorney.
I still have no idea on how to write the paper. It almost made it not worth doing.
It would seem hard to know what to do in the beginning. People tend not to know what they want. And it is a researcher's responsibility to understand their product, their business, and what is technically feasible for them to excel in their market.
Then it is equally hard to make them agree with your decision. Convince them that even though you don't really know what they do, you know what is the best for them, and you know what it takes for them to be successful.
What follows is the easy part, which is the actual research work. Abstracting problems. Reading papers. Designing algorithms. Implementing prototypes. Running experiments. Writing patents. Things people actually expect researchers to do.
The next step would be the boring part. And this includes periodically presenting results and getting feedback. With the only goal of making them take over what you have done. Usually it would mean horrible work like cleaning up code and, god forbid, documentations.
And just before you thought you have arrived to the fun part, namely, writing papers, the worst has yet to come. It is not about writing a high-quality paper. After all, these are industrial research projects with real-world impact that some academic conferences are dying for.
Before disclosing company's trade secret and product's internal details, you need to know if you can actually do that, and if so, how much. In the last few days, in order to decide the position of the paper, I have talked to their engineers, scientist, technical leader, product manager, PR, marketing people, and attorney.
I still have no idea on how to write the paper. It almost made it not worth doing.
Labels: work
2008-02-15
Filming
The big thing in office this week is filming. There is a big sign outside the office building telling us that we should remain silent because of filming. Also, we could be in camera in the building and when it happens we should not look right at the camera. I guess it makes sense to do this here more than in the headquarter in Bay Area, since we have more filming talents (and probably out of work until recently) in LA.
Why filming in the office? They are probably doing some web videos to promote our products. And they seem to think what it takes is some nerdy actors walking around our office building talking with other less nerdy supposedly engineers. And what's the best to mock an office environment than, er..., a real office environment?
Which means that for the last few days, I have been trying to avoid any filming area, to avoid becoming a moving filming background. This has been difficult, since they constantly roam around lobby area. Like yesterday when I left, I had to avoid taking the stairs. They are taking away my only exercise at work!
Why filming in the office? They are probably doing some web videos to promote our products. And they seem to think what it takes is some nerdy actors walking around our office building talking with other less nerdy supposedly engineers. And what's the best to mock an office environment than, er..., a real office environment?
Which means that for the last few days, I have been trying to avoid any filming area, to avoid becoming a moving filming background. This has been difficult, since they constantly roam around lobby area. Like yesterday when I left, I had to avoid taking the stairs. They are taking away my only exercise at work!
Labels: work
2008-02-08
Evil Applications
I've finally got a chance to read my first evil, er... patent application, only five months after submitting my idea. What can I say? It is an interesting experience.
The first thing I noticed was that it is now turned from a small paragraph into a 20-page monster. Then there are several figures of "computer systems" that don't seem to mean much. But most importantly, the text is now rewritten in a language I barely recognize–more than anything, it just seems like a random selection of words to me, consisting of very long sentences with repeating nouns and adjectives.
I was supposed to proofread it. That includes technical correctness and also whether it is as broad as humanly possible. I'm proud to admit that three pages into reading it, I fell into asleep in my cubicle.
There are still two more to come. Anyway, this is a small price to pay, given the few grands I'm getting from each application.
The first thing I noticed was that it is now turned from a small paragraph into a 20-page monster. Then there are several figures of "computer systems" that don't seem to mean much. But most importantly, the text is now rewritten in a language I barely recognize–more than anything, it just seems like a random selection of words to me, consisting of very long sentences with repeating nouns and adjectives.
I was supposed to proofread it. That includes technical correctness and also whether it is as broad as humanly possible. I'm proud to admit that three pages into reading it, I fell into asleep in my cubicle.
There are still two more to come. Anyway, this is a small price to pay, given the few grands I'm getting from each application.
Labels: work
2008-01-24
YAO
It doesn't necessarily stand for a person's family name. It could also mean... YAODBAT, or Yet Another One-Day Bay Area Trip. And I hate these, especially when there are weather issues at both Bay Area and LA airports, making me arriving at LA airport four hours after arriving SFO.
This was for visiting a newly acquired office. It was almost a non-stop meeting from 11 to 5. We have developed a lot of their technologies before the acquisition, and we were there to share with them. And the worst part is, the people there are probably mostly millionaires, unlike me, due to the acquisition...
This was for visiting a newly acquired office. It was almost a non-stop meeting from 11 to 5. We have developed a lot of their technologies before the acquisition, and we were there to share with them. And the worst part is, the people there are probably mostly millionaires, unlike me, due to the acquisition...
Labels: work
2007-10-18
Much Improved Commute
With my office moved from Santa Monica to Culver City, I finally realize, commute does not have to be nasty.
On paper, the distance is shortened only by 30%. But by avoiding crossing two horrible freeways, the commute time is now less than half: from 20-40 mins down to 10-15 mins. More importantly, since the route is not as congested, driving is much more relaxing. No need to worry people's reckless cutting and turning from out of nowhere.
The sad thing is, this is about the only good thing on the move...
On paper, the distance is shortened only by 30%. But by avoiding crossing two horrible freeways, the commute time is now less than half: from 20-40 mins down to 10-15 mins. More importantly, since the route is not as congested, driving is much more relaxing. No need to worry people's reckless cutting and turning from out of nowhere.
The sad thing is, this is about the only good thing on the move...
Labels: work
2007-10-05
24" Commute
Commuting has become much worse lately. That's unavoidable: traffic is only marginally tolerable during the summer, when many people, particularly those go to UCLA, are not around. But it's ok. I've only got one more week to suffer, before my company moves to Culver City, when I can choose to drive a much shorter and better route or just take a bus.
Beside, I have another reason to be excited. I just got a 24" monitor. Since I don't want to repackage it, I haven't opened the box yet. Can't wait to start using it after the move...
Beside, I have another reason to be excited. I just got a 24" monitor. Since I don't want to repackage it, I haven't opened the box yet. Can't wait to start using it after the move...
Labels: work
2007-09-27
Berkeley DB vs SQLite
Ok, this is my attempt to make this blog a little bit more "educational."
In my latest project, like most projects, I need a library to quickly store large amount of data and perform some basic queries on the data. Since I want to implement the prototype quickly and make it easy to send the prototype to other people, I decided against using any "real" database. I see two options: Berkeley DB or SQLite. Since SQLite is like new and cool and allows me to dust my SQL, I decided to use it.
Obviously, the rest of this entry is about why it was a stupid choice.
I need the library to map some ID number and hash values in both direction. Doing it fast in SQL engines requires the use of indices on both fields. In SQLite, the only supported index is BTree, which means that it would needlessly sort the millions of hash values...
A quick reimplementation in Berkeley DB with hash based indexing reduced the running time from 10 hours into 3.
In my latest project, like most projects, I need a library to quickly store large amount of data and perform some basic queries on the data. Since I want to implement the prototype quickly and make it easy to send the prototype to other people, I decided against using any "real" database. I see two options: Berkeley DB or SQLite. Since SQLite is like new and cool and allows me to dust my SQL, I decided to use it.
Obviously, the rest of this entry is about why it was a stupid choice.
I need the library to map some ID number and hash values in both direction. Doing it fast in SQL engines requires the use of indices on both fields. In SQLite, the only supported index is BTree, which means that it would needlessly sort the millions of hash values...
A quick reimplementation in Berkeley DB with hash based indexing reduced the running time from 10 hours into 3.
Labels: work
2007-09-25
Bureaucracy
I never quite understood this word, other than as something opposite to efficiency, until recently. How do a research person define his "business goals"? How about self-evaluation based on alignment of personal performance with vision and value? It is not to the extent of the Colored Giant but at least they rightfully decided that I don't fit in and I agreed.
But things could get worse. Like unwillingly taking precious vacations so that someone other than me can save cost by shutting down sites during holiday seasons...
But things could get worse. Like unwillingly taking precious vacations so that someone other than me can save cost by shutting down sites during holiday seasons...
Labels: work
2007-08-23
Water Park
Free lunch. Free dinner. In between? Free water fun from Raging Waters.
While it was fun at the beginning, after a few attractions, they start look like the same as you are pushed by water from one place to another.
Worse, I got myself burnt. Guess which part. No, it was my left sole. Turns out, walking on bare feet is a fatal mistake, as the sun heat the ground too much...
While it was fun at the beginning, after a few attractions, they start look like the same as you are pushed by water from one place to another.
Worse, I got myself burnt. Guess which part. No, it was my left sole. Turns out, walking on bare feet is a fatal mistake, as the sun heat the ground too much...
Labels: work
2007-08-10
Hate UK Conference Calls
I just did one this morning as part of the evil-turning plan and I didn't enjoy a bit.
First, it has to be in the early mornings due to time difference. Then I have to pay extra attention for the British accent. To make it worse, it is sometimes pretty hard to grasp what they think, since they don't necessarily show their stand or emotion in their tones. Fourth... well I hate to talk to people I don't know well anyway.
First, it has to be in the early mornings due to time difference. Then I have to pay extra attention for the British accent. To make it worse, it is sometimes pretty hard to grasp what they think, since they don't necessarily show their stand or emotion in their tones. Fourth... well I hate to talk to people I don't know well anyway.
Labels: work
2007-08-05
Officially Evil
I declare that I'm officially evil now.
I spent some time this past week learning the patent procedure in my company and started mine. All I can say is, turning evil is a difficult process. First, you need to make some crap that stinks so much that no one has ever touched before. After that, you go and read some worse-than-crap patents that read worse than gibberish (since you know you don't need to understand gibberish). I spent like an hour reading one before deciding that it was something trivial. Then it comes to the worst part, where you actually write your own gibberish. And it wasn't even the application itself--it was only an application for application, subject to approval before the actual application...
But it's not like I have the option of not being evil. Patent applications are a part of my evaluation. Plus, who could resist the path towards evilness, when not only evilness is monetarily rewarded, but also the application to evilness is?
I spent some time this past week learning the patent procedure in my company and started mine. All I can say is, turning evil is a difficult process. First, you need to make some crap that stinks so much that no one has ever touched before. After that, you go and read some worse-than-crap patents that read worse than gibberish (since you know you don't need to understand gibberish). I spent like an hour reading one before deciding that it was something trivial. Then it comes to the worst part, where you actually write your own gibberish. And it wasn't even the application itself--it was only an application for application, subject to approval before the actual application...
But it's not like I have the option of not being evil. Patent applications are a part of my evaluation. Plus, who could resist the path towards evilness, when not only evilness is monetarily rewarded, but also the application to evilness is?
Labels: work
2007-08-02
I Hate Free Food
You heard me right. And I can't believe I said that either. But that's true now.
I don't know who is behind these decisions, but lately, the food catering for meetings are just horrible. They keep ordering cuisines that I don't like, like Mexican, Indian, and Middle East. To make it worse, you don't get to know the cuisine before the meeting day, so it is hard to prepare something else. Why can't we have some more common food, like sandwiches and pizzas?
Man, it looks like I won't have lunch today...
I don't know who is behind these decisions, but lately, the food catering for meetings are just horrible. They keep ordering cuisines that I don't like, like Mexican, Indian, and Middle East. To make it worse, you don't get to know the cuisine before the meeting day, so it is hard to prepare something else. Why can't we have some more common food, like sandwiches and pizzas?
Man, it looks like I won't have lunch today...
2007-07-24
Painfully Debugging
I spent the last two day debugging painfully. It was bad enough that the debugging was done in a beefy 64-bit Linux machine with 8GB of RAM connected via Xwindows tunneled through ssh in a Redhat Linux environment running as a VMware image in my Windows XP laptop. The bug did not emerge until after running the program for half an hour. And the only message I got was a segmentation fault. And debugger only showed corrupted memory...
At last, I realized the bug was because when I compute the memory size for allocation, the result overflowed the 32-bit unsigned long integer. Sigh.
At last, I realized the bug was because when I compute the memory size for allocation, the result overflowed the 32-bit unsigned long integer. Sigh.
Labels: work
