Cloud Computing across LAN's ?

The concept of cloud computing is interesting and actually quite old. It lacked major backing till Google came along and is now increasingly seen as the alternative to PC (given that other alternatives like Tablet PC came and went).

This diagram and definition is from Wikipedia of course ”

Cloud computing refers to computing resources being accessed which are typically owned and operated by a third-party provider on a consolidated basis in Data Center locations. Consumers of cloud computing services purchase computing capacity on-demand and are not generally concerned with the underlying technologies used to achieve the increase in server capability. There are however increasing options for developers that allow for platform services in the cloud where developers do care about the underlying technology.”

What prevents local area networks from enforcing clouds beats me. Put all the apps and ALL the storage on the server.Since most PC OEMS insist on their standard 80 gb hard disk configuration, the IT team of a company has to work harder to enforce it, but once done – They have lower tickets to attend to. Just put thin shell ubuntu PC’s with open office on each local machine. This also makes compliance and productivity tracking much easier to do- just check the server logs. Bottlenecks of course remain that IT Compliance in companies rarely seeks to maximize business value, thus ensuring they are the first to be transferred  to other teams or downsized in downturns as a cost unit not as a core unit.

You can also try Google Apps for enterprise for such initiatives. The software is now ready which wasnt the case a few years back.

Dear Google

Google.com has added its privacy policy to the main page to conform with California law. Here is a question to the masters of the algorithm that I sent to their query system ”

Dear Google ,

I understand that IP Addresses are stored routinely by you, that these IP addresses can be used as unique keys for analytical purposes, but also be used for identifying and locating privacy of people (like in China) with disproportionate technical effort. Why don’t you run a randomizing algorithm that masks the IP addresses but keeps the uniqueness factor alive, and delete the original IP addresses, thus sparing yourself any privacy concerns. The algorithm should be made in a manner that any masked unique  IP number cannot be unmasked , and all same IP addresses have same masked IP addresses.You retain analytical value, consumers retain privacy and we settle this debate once and for all.”
This is in response to its slightly biased privacy policy whose fine print is here ”

http://www.google.co.in/intl/en/privacypolicy.html

Data integrity

Google processes personal information only for the purposes for which it was collected and in accordance with this Policy or any applicable service-specific privacy notice. We review our data collection, storage and processing practices to ensure that we only collect, store and process the personal information needed to provide or improve our services. We take reasonable steps to ensure that the personal information we process is accurate, complete, and current, but we depend on our users to update or correct their personal information whenever necessary.

Accessing and updating personal information

When you use Google services, we make good faith efforts to provide you with access to your personal information and either to correct this data if it is inaccurate or to delete such data at your request if it is not otherwise required to be retained by law or for legitimate business purposes. We ask individual users to identify themselves and the information requested to be accessed, corrected or removed before processing such requests, and we may decline to process requests that are unreasonably repetitive or systematic,  require disproportionate technical effort , jeopardize the privacy of others, or would be extremely impractical (for instance, requests concerning information residing on backup tapes), or for which access is not otherwise required. In any case where we provide information access and correction, we perform this service free of charge, except if doing so would require a disproportionate effort. Some of our services have different procedures to access, correct or delete users’ personal information. We provide the details for these procedures in the specific privacy notices or FAQs for these services.”

This leaves enough loopholes for Google to pick and choose its privacy policy AND its response. Nice spin, but people understanding law, public relations, databases AND algorithms do exist in the non Google world. The New York Times blog “Bits”: is at the forefront. And its a very good blog for all tech news besides the renowned mashable (www.mashable.com) and Silicon Valley Insider (www.alleyinsider.com)

Watch this space.

Online Analytics: Monte Carlo Simulation

Do your eyes glaze over when ever you hear the words simulation ? Simulation refers to trying to predict actual events , usually in a controlled atmosphere. Monte Carlo simulation is a type of simulation that draws on repeated random sampling to hit the result, and it is usually computed using computers (unless you are Enrico Fermi who did it in 1942.)

The classical definition as per the most peer reviewed online statistical journal (called www.wikipedia.org )

“the term describes a large and widely-used class of approaches. However, these approaches tend to follow a particular pattern:

  1. Define a domain of possible inputs.
  2. Generate inputs randomly from the domain, and perform a deterministic computation on them.
  3. Aggregate the results of the individual computations into the final result.”

and ”

Monte Carlo simulation methods are especially useful in studying systems with a large number of coupled degrees of freedom, such as liquids, disordered materials, strongly coupled solids, and cellular structures (see cellular Potts model). More broadly, Monte Carlo methods are useful for modeling phenomena with significant uncertainty in inputs, such as the calculation of risk in business (for its use in the insurance industry, see stochastic modelling). A classic use is for the evaluation of definite integrals, particularly multidimensional integrals with complicated boundary conditions.

Monte Carlo methods in finance are often used to calculate the value of companies, to evaluate investments in projects at corporate level or to evaluate financial derivatives. The Monte Carlo method is intended for financial analysts who want to construct stochastic or probabilistic financial models as opposed to the traditional static and deterministic models.”

Here is a very good example of using the simulation to calculate distribution of leads to a website and the resultant conversion rate and probabilities. The down loadable excel sheet is great for learning both this class of simulation as well , maybe adding more robust mathematics in your online estimation efforts. The site is also great for excel related stuff.mc.GIF http://www.vertex42.com/ExcelArticles/mc/SalesForecast.html 

Use it if you think estimating Online Profitabilty is more complex than :

conversion ratio *number of leads *profit per conversion-number of leads*cost per lead.

The Math behind online strategy games

Online strategy games are the kind people play on Facebook , with thousands of multiple users logging in. The graphics are not as demanding as those in a virtual world (like www.secondlife.com or World of Warcraft http://www.worldofwarcraft.com )

second-life.GIF

These graphical games are hugely entertaining and addictive. unfortunately World of Warcraft is not present either in India or China.

wow.GIF

However this post talks about simpler games (which are more like optimization problems). Here gamers get a specific or variable game currency to spend (which can be tokens, gold ,or dollars). The amounts can be fixed or varying dependent on current position. The gamer then has to allocate the currency into multiple resources that increase his net worth (or prevent his net worth from decreasing by other gamer’s interventions “attacks”) .  These multiple resources can be defensive (to prevent other gamers from stealing worth or value/score) , aggresive (to steal other gamers value) or passively accretive (enhances value on stand alone basis). Usually the costs of these resources are different with different benefits . The benefits are either fixed (like in the www.facebook.com games Triumph or Return of the Infernals ..below) or they vary (as in the game DopeWars Online).

fb_roti.GIF

To further enhance the optimization complexity , gamers can form alliances or cartels to co ordinate strategy against other gamers.

Such games help in simulating operations production, market forces in business strategy (like the beer game) , and have now taken the leap into free online games thanks to Google’s Open social initiative and Facebook’s  pioneering application building by third party developers.

Ultimately they are simply optimization equations which seek to maximize net worth or score of a player subject to multiple fixed and variable constraints.

A good website to track the world of games is www.gamespot.com

This is a fast growing industry and creativity in designing a simple front end,financial resources to host servers , and some maths to run the back end optimization is all that is required .

You might just get some interesting cash flows from advertising, in game promotions,premium subscriptions , and bonus packs besides enjoying the game of course.

Need for Economists in Corporate India

Corporate India has been caught on surprise on many counts recently and most of them are macro economic events.

These have been namely credit rate hikes, inflation due to oil prices (consequent demand for better salaries and attrition) , market entry of new players and above all the rupee appreciation that shave off nearly 1000 basis points off the profitability of unhedged exporters.

Add to this the uncertainty in stock markets over remote events in the sub prime mortgage market in the United States that has actually led to many corporates getting below expectation results in their listing or Initial Public Offerings despite good fundamentals.

All these point to need for better corporate planning and strategizing for economic changes and events especially in a networked world.

Table 1-Top Macro Economic Events that caught corporate India by surprise and their impact

? Credit Policy Hikes by RBI 2006-2007 leading to expensive debt.
? Rupee Appreciation and RBI steps including curbs on ECB.
? Oil Prices and Inflation.
? US Mortgage Market, Effect on Global Equity Markets including India.
? SEZ Policy and impact on communities (this is more of socio-economic topic)

The primary impact of this has been exporters like Infosys missing their earnings guidance due to rupee appreciation, corporates like WNS having lower listed prices ,rising credit costs including for banks , and considerable rework of SEZ plans for corporates like Tatas and Reliance.

These are the biggest names in India, so the impact of lack of econometric planning and forecasting on smaller players is likely to be more.

Most corporates in advanced economies have business intelligence units and economic strategy and planning units. They are used mainly for forecasting sales using scientific quantitative methods like base driver models, time series models and regression models to predict and anticipate demand and align corporate supply and demand chains accordingly.

The usual audience for them is at CXO or Board level advisory positions.

In India while many corporates have started creating these units they are yet to gain the credibility and respect that they would have got in Western Companies.

Main reasons for these are as follows –
depth of Indian academia in application oriented research and their ability to adjust to corporate demands,
skepticism regarding modeling techniques most of which are complex for end users and corporate audiences ,
lack of investment in forecasting soft wares (like SAS , SPSS and even Excel/Solver ) and human resources in these units.

Most Indian corporates would rather hire five more sales managers than invest in two economists who would help create a much better forecast to help plan the corporate strategy.

This is partly due to historic mindsets and partly due to cultural risk aversion, as corporates engage in cost cutting, sales is looked upon as revenue units and planning units are cost centers. An additional complicating factor is that many companies still believe in push based sales, rather than pull based demand targets.

Table 2
Examples of Business Intelligence Units / Planning Units in Indian Corporates.

ICICI
Reliance
Muruguppa Group
Airtel

Examples of Business Intelligence Units / Planning Units in other countries.

General Motors
British Telecom
Nestle
Citigroup

An alternative for corporates unwilling to go into full fledged economics planning units is to become subscribers for customized content providers by third party providers.

This content could be in the form of business research, market research and segmentation studies, predictive models or even economics newsletters. The chief drawback to this is that due to the outsourcing and Knowledge Process Outsourcing boom, sales margins for third party content providers is much more when catering to the global market.

However even for the outsourcing sector it would be advisable to keep a foot in the domestic market, keeping in mind long term growth plans of Indian corporates and the ability to build domain expertise much better while catering to onshore Indian clients rather than offshore global clients. In the short term, these would be lower margins but it would help in building the domain expertise necessary for them to move up the value chain.

As the Indian economy is poised for sustained growth, the size and scale of this domestic demand for economics content would likely scale up manifold. Indian corporates should actually benchmark their demand planning and economic units from international players and partners