Is this the begining of the end of detroit?

Time for new, innovative automakers to step in and solve our problems without the burden of history, pensions, and teamsters in the way.
http://www.engadget.com/2008/06/30/tesla-announces-the-model-s-a-60k-all-electric-five-passenge/

The end of the scientific method: Correlation is More Valuable than Causation?

This great article in Wired discusses the impact of Petabytes of data and our newfound ability to digest it. The basic concept of the article here is that for all time before the world has been full of data that has been unavailable to us, thus, science had to operate with limited information, create theories, gather more limited information, and prove those theories.

Perhaps another approach to science is better. With our newfound ability to gather and process unimaginable amounts of data, why create models and prove them. A better method might be to think of all ways to generate data about a system under investigation, and then allow sheer computing power to find correlations. Notably, correlations on massive amounts of data bear more value than on small data sets.

I see a lot of ideas and value down a path with this style of analysis.. Very cool.

Ecomodding Defined

I mentioned Ecomodding to my little brother who was surprised to hear that you could get such efficiency out of vehicle body modifications. Here's a classic example of ecomodding. It's hideous, of course, but it proves the simple point that relatively simple changes can result in extremely good savings. Thus there is hope for even the humble gasoline engine.

To go along with this article here's some basic math on the cost impact of such efficiency.
My honda civic gets 30-38 mpg and I'm a wasteful aggressive driver. At $4.0x per gallon it costs about $50 to fill the tank up. This ugly car is double the efficiency thus a fill up would be half that (since this ugly ecomod gets ~70 mpg) that's effectively $25 to fill the tank.

So the simple point is companies will make concepts like this sexy, and people will relish the new designs. :o)

Birth of the Spatially Orgainzed Web

It's coming. The technology and bandwidth exists and the first sites are born. Google maps street view and everyscape.com allow the user to virually "walk" through the real world allowing us to map the full wealth of the internet's information onto it. We will see a rapid growth of a spatially organized web which will bring the use of "virtual worlds" into the mainstream.

Here's a video of everyscape's concept.

Don't Buy a Mac!

As a Mac fanboy who doesn't own a Mac (does that even make sense?) I've stumbled upon an obvious and clever idea. If Macs are so happy running Windows then why not run Mac OSx using Vmware in Windows?

All the Mac Fanboys (Fanpeople?) will say that this is ridiculous, but the only reason I don't own a Mac is because of the stupid price tag. My laptop is a 1.6 ghz, 2 gig ram, 100gb hard drive, 15 inch LCD, CD/DVD RW, etc and it costs ~$680 after Tax and shipping (bought 1.5 years ago) and you simply can't get a mac for that price. Minimum is about $1100 bux and they only have 1 mouse button (UG, Steve, we can handle more buttons, I used to have 4...)

So if you want to use OSx and don't wanna pay for a Mac they try using Vmware! :o)

Most Efficient Solar Power

A team at MIT has created a scalable solar power system which could be built and distributed without government subsidies. (MIT site here) This looks great but no commentary is made regarding how electricity generation from this system compares to other solar system. Think about all of the space on tops of buildings around the country. Why not have a few of these on top of each one, tuned to follow the sun to help power the building.

Drill for More Oil? Who's this guy leading our country?

President Bush calls to drill for more oil as some solution to the oil crisis. Last I checked when oil prices were cheap we drove bigger cars farther and never spoke about it; US oil consumption has grown and grown. Now, in light of expensive oil prices we're seeing significant decreases in demand. So why would a marginal increase in the supply of oil, especially without any new domestic refining capacity, really make any difference to oil prices at all? It feels like the US is lead by blind leadership.

Check out this nice little chart of the impact that OPEC has had on the oil prices since they doubled their output. If adding 15,000,000 barrels of oil to the market didn't decrease oil prices then what would our meager 30,000, 0r 300,000 additional barrels do?



Even more humorously, it seems that only Iran's President, Mahmoud Ahmadinejad, seems to have more of a clue than our own leadership...

Your thoughts?

Don't Change your oil every 3000 miles

Save time, save money, save the environment, reduce dependence on oil: Don't change your car's oil every 3000 miles. As discussed in this article, changing your oil every 3000 miles isn't even recommended by the manufacturer of your vehicle, so why do it?!

HS Algebra Applied at Work!

Here's how I applied some basic Linear Regression techniques on the job.

Ok, I have to first say that this one was real fun but I think that means that I'm geeky...

We wanted to see all transactions whose response time standard deviation trended upwards. That sounds simple but it's not quite that simple, especially done in SQL. That analysis requires a basic linear regression which would probably be easily implemented in a statistical analysis tool (SAS, etc) but all I have right now is SQL. Either way, I made it work and it was fun.

The SQL below calculates b1 which is the slope of the line formed by the x,y coordinates of our transactional data. The coordinates are formed as (xs,stddev_1), (xs,stddev_2), (st,sttdev_n), etc. If the b1 is >0 then the slope is positive, if b1<0 b1 =" 0">

So I wrote the SQL below and got all transactions by os whose response time standard deviation trended upwards. This is really interesting too because we get more info than if it trended up, we get a metric on how much it trended up. I've ordered the SQL below by b1 in descending order so we can see what transactions trended up the MOST first! Very cool.

My full data set is attached and a screenshot of some of the data is below.

SELECT  aaaa.os              ,
       aaaa.transaction_name,
       ROUND(aaaa.b1,2) AS b1
FROM
       (SELECT aaa.os              ,
               aaa.transaction_name,
               SUM(numerator)/SUM(denominator) AS b1
       FROM
               (SELECT aa.os                         ,
                       aa.transaction_name           ,
                       b.apdex_dimension             ,
                       aa.xbar                       ,
                       aa.ybar                       ,
                       b.x                           ,
                       b.y                           ,
                       (x-xbar)*(y-ybar) AS numerator,
                       (x-xbar)*(x-xbar) AS denominator
               FROM
                       (SELECT a.os              ,
                               a.transaction_name,
                               AVG(a.x) AS xbar  ,
                               AVG(a.y) AS ybar
                       FROM
                               (SELECT t.os                                                                                                                                                                                                                              ,
                                       t.apdex_dimension                                                                                                                                                                                                                 ,
                                       txn.transaction_name                                                                                                                                                                                                              ,
                                       IF(t.apdex_dimension = 'xs',1,IF(t.apdex_dimension = 'sm',2,IF(t.apdex_dimension = 'md',3,IF(t.apdex_dimension = 'st',4,IF(t.apdex_dimension = 'lg',5,IF(t.apdex_dimension = 'xl',6,IF(t.apdex_dimension = 'xt',7,NULL))))))) AS x,
                                       stddev(trt.response_time)                                                                                                                                                                                                     AS y
                               FROM    test t
                                       LEFT JOIN test_result_transactions trt
                                       ON      t.id = trt.test_id
                                       LEFT JOIN transaction txn
                                       ON      trt.transaction_id = txn.transaction_id
                               WHERE   t.project                  = 'as 8.0 sp3 pvt'
                                   AND t.test_type                = 'apdex'
                               GROUP BY t.os                ,
                                       t.apdex_dimension    ,
                                       txn.transaction_name ,
                                       IF(t.apdex_dimension = 'xs',1,IF(t.apdex_dimension = 'sm',2,IF(t.apdex_dimension = 'md',3,IF(t.apdex_dimension = 'st',4,IF(t.apdex_dimension = 'lg',5,IF(t.apdex_dimension = 'xl',6,IF(t.apdex_dimension = 'xt',7,NULL)))))))
                               ) a
                       GROUP BY a.os ,
                               a.transaction_name
                       ) aa
                       LEFT JOIN
                               (SELECT t.os                                                                                                                                                                                                                              ,
                                       t.apdex_dimension                                                                                                                                                                                                                 ,
                                       txn.transaction_name                                                                                                                                                                                                              ,
                                       IF(t.apdex_dimension = 'xs',1,IF(t.apdex_dimension = 'sm',2,IF(t.apdex_dimension = 'md',3,IF(t.apdex_dimension = 'st',4,IF(t.apdex_dimension = 'lg',5,IF(t.apdex_dimension = 'xl',6,IF(t.apdex_dimension = 'xt',7,NULL))))))) AS x,
                                       stddev(trt.response_time)                                                                                                                                                                                                     AS y
                               FROM    test t
                                       LEFT JOIN test_result_transactions trt
                                       ON      t.id = trt.test_id
                                       LEFT JOIN transaction txn
                                       ON      trt.transaction_id = txn.transaction_id
                               WHERE   t.project                  = 'as 8.0 sp3 pvt'
                                   AND t.test_type                = 'apdex'
                               GROUP BY t.os                ,
                                       t.apdex_dimension    ,
                                       txn.transaction_name ,
                                       IF(t.apdex_dimension = 'xs',1,IF(t.apdex_dimension = 'sm',2,IF(t.apdex_dimension = 'md',3,IF(t.apdex_dimension = 'st',4,IF(t.apdex_dimension = 'lg',5,IF(t.apdex_dimension = 'xl',6,IF(t.apdex_dimension = 'xt',7,NULL)))))))
                               ) b ON aa.os                 = b.os
                           AND aa.transaction_name          = b.transaction_name
               ) aaa
       GROUP BY aaa.os,
               aaa.transaction_name
       ) aaaa
WHERE   aaaa.b1 >0
ORDER BY aaaa.b1 DESC