Anti-reservation

With all the reservation controversy in India, I felt I had to express myself too. So, here goes :

I think that the goal of any affirmative action program should be to redress the
differences in oppurtunites and access to resources that people have in life.
In modern day india, caste is no longer a clean and clear dividing line.
i.e. there are "lower caste" people who are very well-to-do and at the same time
there are "upper caste" people who are not well off.

So, any affirmative action program that is caste based is going to have real problems
actually benefiting people who need to be helped. I think that an affirmative action
program that is grounded on economic status is something that will help people
individually and the country as a whole a lot more. Ideally, such a program would
have a graduated scale rather than all-or-nothing, providing more help to people who are
worse off (as a side benefit, reducing incentive to cheat by getting false certificates)

In my experience, the current reservation system mainly helps lower-caste
students get ahead of better qualified upper-caste students of similar economic
means and opportunities. On the other hand, a graduated program based on
economic circumstances will actually benefit us all by allowing the most talented
youth to come forward in spite of the problems they may have suffered in life.

Also see Bay Area 4 Equality

American Business 1920-2000

I just finished reading "American Business 1920-2000"
Being the first history book that I have read in a long time, I was kind of expecting it to be a dull compendium of facts. Thankfully, it is not like that at all. The author has divided the time period into 5 sections, and for each section he picks a core theme and talks mainly about a few companies around that theme. Each theme is the industry that came of age in that period and had a large impact on american society. Interesting interleaved between the major theme chapters are minor digressions into other interesting topic : women / blacks in business; the financial industry, the chemicals industry etc.

My core takeaways from the book :

1920s - Motor Vehicles and Modern management
- The evolution of the car industry struck me as having a lot of similarity with the computer industry of the 80s-90s. Lots of innovation at the start. Constantly dropping prices. Evolution to a more stable industry. I had not realised the generality of this structure.
- I was impressed with Alfred Sloan's management insight into decentralization. Employee empowerment to the right extent was the key to GM's success. And it would benefit us all to pay more attention to the influence of corporate structure on success.

1930's - P&G
- Brands are important and should be treated as such.
- Consumer research is a powerful tool.

The New Deal and WWII
- Interestingly, the regulation of the financial markets provided by the new deal turned out to be beneficial for the financial markets themselves by making them more accessible. Market forces would not have moved in this direction, because market forces end up taking a short term view.
- Ferdinand Eberstadt's key insight was that it is only the limiting resource in the whole system that needs to be regulated centrally. All the other resources would line up. Similar to Marv Burkett's point that for resource allocation, expenses should be measured in terms of the most constrained resource for the company, which may or may not be money (e.g. it could be number of employees).

RCA
- More of a description of what went wrong than what was done right.

McDonalds
- Franchising is essentially the sale of intellectual property and goodwill (in the form of a brand). I had not realised this so clearly.

Overall
I can't find where I read it in this book, but I found the statistic that, the gap between the rich and the poor has been growing for the last 30 years, after declining for the 50 (70?) before, to be kind of disturbing. I feel that an excessive gap between the rich and the poor is bad in 2 ways. One is that it promotes social unrest if it rises too high (French revolution). Or at least social dissatisfaction because not everyone has every opportunity available to them. Another more subtle effect is that it reduces competition in anything requiring a large investment, which means that society doesn't get the most capable people in those important positions. We can already see that most politicians in the US are people who are already wealthy.

Chapter 2 : Through the pipeline

Real-Time Rendering :
This chapter discusses the pipeline architecture used in current real-time rendering systems (PCs, game consoles). Broadly, there are 3 stages : application, geometry & rendering. In more detail :
The application stage executes in SW and feeds the geometry stage. It handles user interaction. It also deals with collision detection. Other stuff includes "texture animation, animations via transforms, geometry morphing, or any kind of calculations that are not performed in any other stages".
The geometry stage is broken up into :

• Transforms : Converting from model coordinates to world coordinates to eye space.

• Lighting & Shading : Texturing, color interpolation, lighting.

• Projection : There are 2 types - Orthographic & Perspective. Orthographic transforms the view box to the unit cube. Perspective transforms the view frustum to the unit cube. (The unit cube has edges of size 2, center at the origin)

• Clipping : Cull primitives that are partially or completely outside the unit cube to save work.
• Screen Mapping : Shift to screen coordinates.

The rasterizer stage is described as one unit. This stage handles the conversion of primitives to pixels. That includes Z-occlusion, alpha blending, operations using stencil buffers.

My only nit with this chapter was the stylistic differences between the geometry section and the rasterizer section.

Real-Time Rendering

I am going to read the book on Real-Time Rendering by Akenine-Moller and Haines.
And to make myself read it properly, I'll try to summarize and review the chapters here, as I finish them.

Okay, so here goes :
Chapter 1 : Introduction
Summary
Real-Time Rendering is defined as 3-dimensional interactive (10+ fps) rendering. Then there is a contents and notation review. The notation review would have been better if there was some explanation of the maths used in the illustrative examples. I had to go read the appendix on homogenous notation and think a while about the plane equation and atan2 (converts direction to angle). Not a big problem, though.

Harry Potter

With the new Harry Potter book coming out, I realised that I was mighty confused about what had happened in the previous ones.

So I found some cool links with a summary of those books :
Book 1 : Philosoper's Stone
Book 2 : Chamber of Secrets
Book 3 : Prisoner of Azkaban
Book 4 : Goblet of Fire
Book 5 : Order of the Phoenix

Very useful.

A cool new tool called subtext. It claims to be a language, but it is more like excel to the nth degreee without the grid layout. Interesting nontheless for some application fields.

LUT
Subtext
Neat (but long) demo

optimal design

It seems that it is possible to optimize any problem that can be formulated as an LP.
It is also possible to solve discrete problems the same way by writing them using continuous variables with appropriate constraints and noticing that optimal points are always are the vertices of the feasible space.

Is it possible to come up with optimal logic design(, or optimal design of other sorts).
One way to do this would be to come up with some way to constrain the design space.
But there are potentially unlimited ways to even generate a circuit as simple as an inverter.
The tricky part that I need to figure out is how to constrain the design space to junk the stupid stuff without junking the unusual but potentially good stuff.