Yep. You hit it right on the head. Google isn't a search/data organization company. They are an advertising company. All of their products are designed to do one thing: sell advertisements. In that business model, they have found a way to compete very aggresively with other software companies because most software companies rely upon sales of software for their income. With Google's business model, they care only about selling advertisements, so they can give their software away for free. What does that mean? Nobody can beat them on a cost basis. Other companies can produce higher quality (or more feature-rich, in MSFT's case) software, but nobody can beat them on cost. Over time, Google can continue to improve their apps until they approach the quality of traditional software. What this means is that Google can chip away, and eventually eclipse, other software manufacturers in terms of quality and features, but nobody can eclipse Google on cost.
Another interesting thing about Google's model is that, compared to traditional Madison Street advertising companies, most of Google's revenues come from small to medium-sized businesses. They've levelled the playing field when it comes to buying advertising space, allowing a mom-and-pop shop to compete directly with a mega-conglomerate. What I find most interesting about this model is that I believe it to be fairly immune to business cycles. While large companies will expand and contract their advertising budgets based upon their bottom line, Google will receive relatively constant business from the mom-and-pop shops, whose advertising budget is both small and fairly constant regardless of recessions or expansions. We'll have to see how Google does through the next recession, since during the last recession they were still growing market share far faster than the economy was contracting. My bet is that Google becomes a safe haven for investors during recessions as a result.
Finally, despite Google's phenomenally high P-E ratio, Google is currently fairly well valued, at least according to Free Cash Flow models. According to my research (and please don't take my word for it, do your own research--the results I found were startling to say the least), the FCF model was the only model that was significantly better than a random walk in predicting company valuations. Google has a high P-E ratio not because it is overvalued. They have a high P-E ratio because they have fantastic (and improving!) profit margins and revenue growth. Can it continue? Well, past performance is certainly no guarantee of future performance, but based on what I perceive to be a business model with a clear and sustainable competitive advantage, and a relatively non-existant connection to recession/exapnsion, I believe they will be able to sustain strong revenue growth and stable or improving profit margins for the next 3 years at the least. Plug in those assumptions to a FCF model, and you'll find that GOOG is fairly valued, if not undervalued.
Yeah, that is the truth. My comment about the NSA rooms is that, after reading the stories last year, the room I was in matched the descriptions perfectly. Palm scanner, retinal scanner, and serious security. Oh, and there were tons of Juniper fiber routers in there. And that was in 1999!
I once worked for a dot-bomb e-commerce company. We had a product that tied into several major credit card issuers (i.e. >40% worldwide market share for issued credit cards). As part of the installation and maintenance of the product, I got to spend many weeks in MAE East (perhaps the biggest data center in the world). From what I've read about the Baby Bells' special networking rooms when the NSA scandal broke last year, I wouldn't be surprised if these servers shared one of those special rooms with the NSA routers.
The data center was about 5 floors below ground level. No form of wireless communications worked whatsoever--cell phones, pagers, etc. Once I parked my car, I had to go to an unlabeled metal door with a tiny camera on the top. Security guards would buzz me in and require me to sign in at their station. Then I would get buzzed in to the main data center room that contained another room inside of it. From there, I had to enter a password into another security system and place my palm on a palm scanner. Inside this room was another security guard--I would have to sign in with them, too. Then I would enter a different password into another security system, and place my head in front of this retinal scanner. This would buzz me into another room with the cages for each of the clients. There was a padlock on the cage, behind which were our servers. The servers required two separate smart IDs to be placed into an external card reader so that there had to be at least 2 people there to perform any maintenance. The servers themselves were locked down pretty tightly, too. It all seemed pretty insane as far as security goes, but I understood--these computers contained every credit card for the credit card issuer.
Well, after about 3 days of going to this data center, everyone got to know me. They would sign in for me to speed up the process. The security guard behind the door with the palm scanner used to get very hot, so she would often block the door open, thus defeating the palm scanner. The retinal scanner also had problems, often requiring about 3 tries before it would read correctly, so that door was often blocked open, too. Then, one day one of us had forgotten our smart card. We started cursing, as the round trip to pick up the card was about 45 minutes, so we tried it with only one smart card. Bingo. It worked. So then we tried it with no card. Seems the card readers weren't functioning properly. So, overall, we were able to defeat all of the security measures except for the padlock, and all because the security staff (getting paid 2 bucks above minimum wage, no doubt) all "knew" us. In my humble opinion, it would have been far smarter to *not* have the security guard in the foyer behind the palm scanner. After all, social engineering is probably the most common form of circumventing security.
Another funny thing about this was that we had a rather difficult security audit for all code releases. We had a bunch of ex-NSA employees working for us that were rather good about it, too. We would also hire outside auditors to do reviews of major code releases. It was all fantastastic, except for one thing: code patches didn't get the same scrutiny as code releases. In fact, they got none. Well, in order to expedite the release of one particular feature (that required emailing confirmation to customers), we packaged it as a "patch". No security audits. And for something that required the installation of a mail server! Furthermore, the code base had access to the record-level encryption used to store the credit cards. So, basically, if I had wanted, I could have installed a bit of code that would have decrypted all of the credit cards of users of our software and emailed them to a third party. I could not believe it. It's a good thing I have what I consider to be high moral and ethical standards.
I realized through this ordeal that security measures are not put in place to ensure security. They are put in place to give people the perception of security. And, furthermore, automation and removing the human element are good things for security. People should be used to monitor and oversee automated security, not to be actively involved in that automated security.
One word: immigration. The US still has far and away the best universities in the world. There are a lot of reasons for this, but probably the biggest one is that professors can make several times the money in the US as they can in the UK or in mainland Europe. Additionally, there is a great deal more investment capital for startups in the US, encouraging many of the brightest to start companies in the US. As a result, many of the best European researchers come to the US for their research, and stay here to start high-tech companies. This is not a new phenomenon; simply look at some of the most famous US scientists over the years to see how many of them are immigrants (such as Alexander Graham Bell and Albert Einstein).
Last I checked, something like 60% of the graduate students at these US universities are from other countries. A significant number of those stay after achieving their degrees. Even though the primary education system in the US falls short of many (most?) developed countries, the US has succeeded in attracting these foreigners, with many of them immigrating after achieving their degrees.
The biggest threat to this cycle? US foreign policy and immigration policy. With this "global war on terror", the US has made immigration far more difficult. At the same time, the image of the US has suffered tremendously as a result of US foreign policy, trade policy, environmental policy, and human rights policy. If the US stops this cycle of immigration as a result of its self-centered governmental policies, then, yes, the US very well may lose its edge in the global economy.
Yes. After all, it's so difficult from the naming convention to tell that the Solo has one core, and the Duo has two cores. Yes, the naming convention is pretty confusing. With a name like Turion, it's so obvious that it has only one core. (OK, so I'm ignoring the Core 2 Duo name...that one *is* pretty f'ed).
My concept of "fair value" for a corporation is the net present value of their future free cash flow, plus their assets, minus their liabilities.
that's not what the grandparentposter said. He said that market cap is not a good indicator of how impactful a fine is. For example, it might affect existing credits adversely, increasing the cost of existing lines of credit as their credit rating drops. That's an effect unrelated to market cap, and related more to things like the actual assets and cashflow.
So, since the fine affects the credit [rating] adversely, according to you, this affects their cost of credit, and therefore impacts cash flow. Well, lo and behold, my concept of "fair value" says the market capitalization should drop to reflect their cash flow. So, the market cap *is* affected. In fact, my basic argument is that the market capitalization becomes a good proxy for the impact of the fine.
Anything and everything can and will influence the stock quotes themselves, and influence market cap, including whether the weather is nice or not, if some trader is having a good day, etc.
And if the market cap (i.e. share price) gets out of whack because a trader accidently puts in a sell order for 10M shares instead of 10 shares, then this will present an arbitrage opportunity. Arbiteurs look for these events and *will* correct this. Arbitrage does fix this, although the fix isn't instantaneous (as those who believe in perfectly efficient markets would have you believe). How fast it gets fixed is a function of the liquidity of the market. The more liquid the market (i.e. the greater the # of buyers and sellers, and the greater the volume of trading), the faster the correction will occur.
I'll give an example where this simply won't work. Consider the largest owner of commercial real estate (shops) in, say, The Netherlands (Maxeda). Maxeda's assets, its real estate, are worth more combined than its marketcap! Madness, you'd say! The invisible hand of the market should correct this outrage! Either Maxeda is worth more, or a corporate raider should split the company up and sell it in parts.
The value of the corporation is not a function of thier net assets alone (also known as their book value). As I said before, the FCF valuation method shows that it is a funciton of their net assets plus the net present value of their future free cash flows. If a company is trading below book value, then I can fathom a number of reasons this might be the case. First, the accounting value of the assets may be incorrect (this is basically what you are saying regarding Maxeda...splitting the company up & selling it off won't return the accounted value of the assets, because the accounting value is greater than the liquidation value). Second, the company might be hemorrhaging free cash flow. If an arbiteur sees this second case (as Kirk Kerkorian does with GM), the arbiteur might decide to take a large minority interest and change the management so that it won't hemorrhage cash (as Kerkorian is doing with GM), or the arbiteur might decide to split up the company & sell it off. In either case, the goal is to prevent erosion of the asset value that results from negative cash flows.
The same goes for market cap, as the grandparentposter eloquently explained; if you suddenly dump every share on the market, you won't get a good price for them - and it's doubtful the price would recover. Conversely, if you try to do a hostile takeover of a company, expect to pay a premium over the market value, like Mittal recently did to acquire Arcelor. There was no magic arbitrage going on there that made sure Mittal paid the right price; in fact, they probably paid too much, but they expect to make good on that by using their controlling interest to do new and exciting things. And that is merely a gamble.
In the case of Mittal and Arcelor, the Mittal family believes that acquiring Arcelor presents opportunities for new future cash flows that would not be poss
What about arbitrage? Basically, if a stock goes into freefall for no apparent reason other than a change in supply/demand for stock trades, then the value of their shares would be lower/higher than the value of the company. This is a classic case of arbitrage, and will eventually lead to an arbiteur to buy (if it is undervalued) or short (if it is overvalued) the stock. If the company is significantly undervalued, an arbiteur might decide to purchase a significant portion of the outstanding shares and attempt to divide the company into pieces & sell it off. This is what a corporate raider does.
What is a fair value for a company's shares? This is a subject of great debate. However, based on a statistical analysis I did comparing a slew of the more common ways of valuation for a company, I found that the most accurate valuation method is FCF (Free Cash Flow)--basically, count up all of the company's future free cash flow (i.e. sort of like profits, but based on true cash flows, not on accounting profit) discounted for inflation, and subtract the company's net liabilities. A one-time fine will reduce the value of the corporation by the amount of the fine plus any perceived impact the fine will have on future free cash flows (i.e. the fine will likely hurt the image of the company, and therefore reduce future sales), and will impact the proper valuation for the company. So far this morning, since this fine was announced, the price of the stock has dropped $0.358. Multiply this by the number of outstanding shares (10.20 billion), and you get a financial impact of something like $3.65 billion. Of course, this isn't the only thing impacting the company's business (a 20% chance of Vista not shipping also plays a part). But, to say that Microsoft's market capitalization is not affected by this ruling is naive at best, and outright wrong at worst.
Supply and demand do have short term impacts on share price. But, in the long run, the arbiteurs will restore the share price back to normalcy. The more liquid the market, the quicker this process of restoration will occur. Market capitalization is not a joke number, since the owners of a company's stock get to vote to control/guide the corporation.
Oh, man, I think I had better buy some silicon dioxide futures then. Or, I suppose, we could always just go to a desert & pick up some sand...
Kidding aside, manufacturing cost curves for solid state follow the First Corrolary to Moore's Law--that the price per transistor halves every 18 months (Moore's Law states that the # of transisters doubles every 18 months, but the corrolary also proves true). According to this article, the period for photovoltaic cells is about 4-6 years. Hard disk drives seem to be following a period of 2-5 years, with a greater deal of variance than with solid state. It is inevitable that solid state will be cheaper than hard disk drives at some point.
The same holds true for LCD panel monitors. Some day, wall-sized LCD panel monitors will be only slightly more expensive than wallpaper, and we'll be living in a Total Recall world.
Back about 15 years ago, I calculated the crossover point for solid state vs. hard disks as occuring in 2008. I don't think I'll be too far off...maybe 2010?
Great question, especially considering Sun now offers N1 for free. I guess the massive amounts of configuration & app support required for it still make it expensive, and Sun is hoping to get some of that work in the form of a consulting contract.
This question, far more than one about the absolute cost of Cassatt's solution, is the one that will decide whether I choose Sun N1, Cassatt, or whether I choose to build out my data centers so that each app has dedicated hardware. After all, the $5k per server costs would cover probably 1-3 months of the total costs in the real world of running a server; if I can eliminate those costs, $5k/year is a bargain.
Try cooling a dual proc computer without the case sounding like a jet engine...better to go with dual core. At least for the type of procs that can decode HDTV
$2276 is for a dual HDTV tuner setup. The case is so that it looks pretty and doesn't sound like a jet engine. The $300+ cpu gets them a dual core cpu (see below).
I'll agree on the RAM, although if the computer is used solely for a media center PC, 2GB of memory isn't that helpful. And I'll also agree on the hard drive, having just purchased a $200 400GB hard drive myself for my media box.
Honestly, given the proper motherboard, onboard sound with digital outputs going to the DAC in my stereo, and I see no need for a $100 sound card.
Just try to decode an HDTV signal on a $150 CPU. It can be done, but not if the PC is doing anything else. They went dual core, I would bet, so that they can decode an HDTV signal while simultaneously recording 2 HDTV signals and a regular signal, all while doing pretty much anything else without skipping a beat.
It's not perfect, but it is fairly reasonable. It does feel like they picked the components of their advertisers, though.
If a "spaceship" is travelling at, say,.95c, then turns on a flashlight pointed in the direction of travel, a casual observer in front of them would see the light travelling at 1.0c, and the "spaceship" travelling at.95c. To someone on the spaceship, the light will be leaving the spaceship at 1.0c.
How is that possible? Time/space dialation. In effect, time appears to slow down for an observer on the spaceship (relative to the person on the ground). Additionally, space appears to lengthen for the observer on the spaceship, so that the distance between the spaceship and the "fixed" appears longer. The net effect? The observer sees the light leave the spacecraft at 1.0c, and the fixed observer sees the light separate from the spaceship at.05c. The wavelength of light will appear different for each observer, due to the differing apparent distances and timeframes for the two observers.
Actually, Moore's Law relates to the number of transistors per unit area, not processing power or clock speed. The corrolary to Moore's Law indicates that the price per transistor halves every 18 months. Nothing is ever said about clock speed or processing power. Check this link for the actual wording. By the way, Moore's Law still seems to be holding, and with the proliferation of multi-core computing and whatnot, I expect Moore's Law will hold for a while to come. The difference now is that we will start to make better use of multiple layers on a single piece of silicon, instead of working towards smaller transistors. We are, indeed, starting to bump up against the laws of physics with the size of transistors, but we're only starting to explore the benefits of creative packaging.
This is not a case of air not being linear. This is a case of the summation of wavelets resulting in beaming, as opposed to a radial dispersion pattern. This is true of any wave (electromagnetic, water, or air/gas). If the width of the transducer is less than 1/2 wavelength, you get a radial dispersion pattern. Likewise, if the tranducer is greater than 1/2 wavelength, you get no dispersion (hence beaming). This is the reason why speakers have enclosures. The enclosure prevents the backwave from cancelling with the front wave. This is also the reason why eclosureless speakers (like those from Martin Logan and Magnepan) tend to be fairly large.
The combination of frequencies is known as aliasing. Aliasing is what allows sonic frequences to be created from two supersonic frequencies.
There's nothing about linear versus nonlinear behavior of air in effect there as far as I know.
One advantage of this effect, for audiophiles (at least once the patents expire--I believe they were awarded in the mid-90s), is that a single tranducer can reproduce then entire audio spectrum. 20Hz-20kHz is 8 octaves, a mammoth feat for any single tranducer to reproduce in linear fashion (by linear, here, I mean that doubling the input power doubles the output power). However, move that from 100kHz-120kHz (resulting in the same 20-20K range when using this heterodyne concept), and the tranducer only needs to create roughly 1/4 of an octave linearly in order to recreate the entire audio spectrum. That task is far easier to do than doing it the "old fashioned" way.
In about 10 years, I would expect to see a plethora of audiophile-quality speakers that consist of an incredibly tiny flat square panel tranducer, coupled with a reflecting device that reflects the sound in a radial pattern. Or perhaps we'll end up with directional reflectors that are able to follow us, allowing a "Minority Report"-type directional advertisements.
What exactly does heterodyning have to do with air molecules behaving nonlinearly? I understand acoustics fairly well, and understand that if the baffle around the transducer is greater than 1/2 wavelength, you get beaming instead of radial dispersion. This, combined with aliasing effects, is what I understood caused heterodyning. I've never heard anything about air molecules behaving nonlinearly as a cause. Can you elaborate?
Your apparent description of the problem (heavy swapping, a GC that runs at low priority) is simply wrong. The problem you were seeing is that there was no available space in the "new" heap space for new objects, so objects were immediately promoted to tenured space. Once the tenured space gets to a certain point (I think the default is 75% full), the tenured GC runs and stops the JVM cold in its tracks (if a naive GC algorithm is used). If there is no space to be reclaimed during a tenured GC, it can take a really long time to run. If your heap is large (>512MB, in my experience), this can take >10 seconds.
Use the "concurrent" GC algorithm, and these "stop the world" delays are reduced to less than 0.2 seconds, in my experience.
To use the concurrent GC algorithm, start the JVM with this command-line option: "-XX:+UseConMarkSweepGC". Also, you probably want to do some analysis of the JVM churn of memory. There is a good chance your application may require more "new" space, or different ratios between new & old spaces, and for the survivor space.
If you are having problems with swapping, buy more memory. I don't think porting the code to C++ will help.
A better solution to your problem is to quit doing XSL transforms, XPath, or DOM processing of XML documents in a server-side application. The app where you saw these delays did heavy XML processing, right? Yeah, I thought so. Try using a SAX parser, and re-implement your transforms to work on small blocks rather than entire documents. The problem is you are creating too many new objects (something that XML processing does a lot of) without freeing them, causing your "young" space to get filled up before the "young" GC can run. This causes the tenured space to get cluttered with "new" objects, resulting in fragmentation and increased frequency of tenured GCs. That, in turn, causes a "stop the world" response that compounds the problem, since the JVM is spending >30% of its time attempting tenured GCs when there is nothing it can garbage collect (since it is filled with "new" references).
For most non-American speakers of English (as well as many Americans), the Oxford English Dictionary is the reference for the spelling of English words. Well, the OED lists the -ize spellings as the primary spellings for most words (but not analyze/analyse). Here is a link for you.
On a related note, the word 'soccer' was introduced by the British, as well. It seems that the move towards "-ise" and football are attempts by the Brits to appear more like the French.
Interesting. That's not what Standard & Poor's say. From their website:
An index consisting of 500 stocks chosen for market size, liquidity and industry group representation, among other factors. The S&P 500 is designed to be a leading indicator of U.S. equities, and it is meant to reflect the risk/return characteristics of the large-cap universe.
Companies included in the index are selected by the S&P Index Committee, which is a team of analysts and economists at Standard and Poor's. The S&P 500 is a market-value weighted index, which means each stock's weight in the index is proportionate to its market value.
Sounds like it's not a far cry from "500 companies we think are cool".
This approach of measuring corporate valuations is called FCF (Free Cash Flows), and should be based upon free cash flows, not on dividends. It is very commonly used in the investment community, and shows a *much* higher confidence factor when compared to actual valuations than either trailing or leading P/E valuations.
Looking at http://finance.yahoo.com/q/ae?s=GOOG we can see that Google currently has a trailing P/E of about 121, but a forward P/E of around 46. That forward P/E really is not too outrageous for a company in a high-growth phase.
Take a look at the FCF calculations and you can see that Google may even be undervalued, depending on your estimates of future growth and profitability. Google has shown fairly consistent gross profit margins, and improving operating profit margins as they have grown. Furthermore, they are showing a growth rate of >40% per year in revenue. If they can maintain that for 3 years, then reach a steady state in growth & profit margins, their $300+ stock price is actually fairly valued. If they can exceed that growth, sustain the growth for longer than 3 years, or improve their profit margins, then an even higher stock price is warranted. Remember, stock valuations aren't based upon what the company did in the past, but what investors think will happen in the future.
Given that, I think everyone here has missed the boat. Their search engine is wonderful, and drives a lot of their business. But remember that ad sales from their search engine accounts for only about 1/2 of their revenue. The other 1/2 comes from Ad-sense selling to other websites. If Microsoft releases a magical search engine that manages to steal 100% of Google's search engine business, Google still would only lose 1/2 of their revenue. Based on their profit margins, I would bet they would be able to remain profitable in the event of such a change, too.
Google is an advertising company. They make money by driving traffic to their (free) web sites, and by selling their superior ad technology to other websites. For all that everyone may hold Google's search engine technology in high esteem, their targeted advertising is equally impressive. Combine this with the multitude of data streams they have to collect information about the visitors to theirs and other websites that use AdSense, and they have a killer product that cannot be matched easily by any of their competitors.
$300 is a fair valuation. Anyone using P/E ratios to demonstrate that a company is either over- or under-valued is, imho, a moron. P/E ratios don't capture growth rates, and don't reflect the amount of free cash flow a company can generate. In the end, it is free cash flow (and not profits) that drive a company's worth. It is very difficult to "cook the books" with respect to cash flows, also, so FCF valuations will be more immune to Enron-esque bookkeeping, too. Don't believe me? Look at the FCF generated by Enron in 1996-2000. The writing was on the wall then. And, the writing is on the wall now with Google. Google is a sound company with a killer business plan. Furthermore, they have an excellent record at execution that I believe indicates they will be able to sustain their growth for a long time to come.
Damn, no mod points. I didn't laugh at the original (grandparent) comment, but yours had me in stiches. Who says you can't convey sarcasm in written form?
So if you are a haliburton employee and you are against the war in iraq you should be fired?
If you throw in speaking about it in public (and I think the grandparent post implied that), then you've got a situation where a person is "publicly stating an opinion that is contrary to the business interests of the corporation". Is that acceptable to fire someone that voices such an opinion? I don't think so. I think the grandparent was right on target with their statement.
However, if the person was working in a right-to-work state in the U.S., then I believe the employer has the right to fire that person for any reason so long as it is not discriminitory. I don't believe this would count as discriminitory. I believe this argument was used as the justification for the private corporation that fired an employee during the U.S. presidential election primaries because the person attended a Bush rally wearing anti-Bush shirts.
Re:Is it just me, or why not explain it better?
on
Eclipse 3.1 Released
·
· Score: 1
Man, you are a Java developer and you've never heard of Eclipse? Where have you been hiding? Let me guess--you've never heard of SWT, either?
If you are a Java developer and you haven't heard of eclipse, download it. It's the open-source IDE originally from IBM. It seems to be the IDE with the most traction in the Java world. IIRC, even Borland and BEA have decided the next versions of their respective IDEs (JBuilder and Workspace) will be simple plugins to Eclipse.
It's quite a nice IDE. Really. There's a ton of useful plugins for it, too. I'm not so sure it's any better, technically, than NetBeans, but it sure has a critical mass of 3rd party support.
Yes, the Jetty lead works for Gluecode now, I think. That's one of the reasons the JBoss/Jetty integration has been dropped. There's bad blood between JBoss & Geronimo.
Slashdot Mirror
Yep. You hit it right on the head. Google isn't a search/data organization company. They are an advertising company. All of their products are designed to do one thing: sell advertisements. In that business model, they have found a way to compete very aggresively with other software companies because most software companies rely upon sales of software for their income. With Google's business model, they care only about selling advertisements, so they can give their software away for free. What does that mean? Nobody can beat them on a cost basis. Other companies can produce higher quality (or more feature-rich, in MSFT's case) software, but nobody can beat them on cost. Over time, Google can continue to improve their apps until they approach the quality of traditional software. What this means is that Google can chip away, and eventually eclipse, other software manufacturers in terms of quality and features, but nobody can eclipse Google on cost.
Another interesting thing about Google's model is that, compared to traditional Madison Street advertising companies, most of Google's revenues come from small to medium-sized businesses. They've levelled the playing field when it comes to buying advertising space, allowing a mom-and-pop shop to compete directly with a mega-conglomerate. What I find most interesting about this model is that I believe it to be fairly immune to business cycles. While large companies will expand and contract their advertising budgets based upon their bottom line, Google will receive relatively constant business from the mom-and-pop shops, whose advertising budget is both small and fairly constant regardless of recessions or expansions. We'll have to see how Google does through the next recession, since during the last recession they were still growing market share far faster than the economy was contracting. My bet is that Google becomes a safe haven for investors during recessions as a result.
Finally, despite Google's phenomenally high P-E ratio, Google is currently fairly well valued, at least according to Free Cash Flow models. According to my research (and please don't take my word for it, do your own research--the results I found were startling to say the least), the FCF model was the only model that was significantly better than a random walk in predicting company valuations. Google has a high P-E ratio not because it is overvalued. They have a high P-E ratio because they have fantastic (and improving!) profit margins and revenue growth. Can it continue? Well, past performance is certainly no guarantee of future performance, but based on what I perceive to be a business model with a clear and sustainable competitive advantage, and a relatively non-existant connection to recession/exapnsion, I believe they will be able to sustain strong revenue growth and stable or improving profit margins for the next 3 years at the least. Plug in those assumptions to a FCF model, and you'll find that GOOG is fairly valued, if not undervalued.
That's my 1.5 cents (inflation is a bitch).
-dan
Yeah, that is the truth. My comment about the NSA rooms is that, after reading the stories last year, the room I was in matched the descriptions perfectly. Palm scanner, retinal scanner, and serious security. Oh, and there were tons of Juniper fiber routers in there. And that was in 1999!
I once worked for a dot-bomb e-commerce company. We had a product that tied into several major credit card issuers (i.e. >40% worldwide market share for issued credit cards). As part of the installation and maintenance of the product, I got to spend many weeks in MAE East (perhaps the biggest data center in the world). From what I've read about the Baby Bells' special networking rooms when the NSA scandal broke last year, I wouldn't be surprised if these servers shared one of those special rooms with the NSA routers.
The data center was about 5 floors below ground level. No form of wireless communications worked whatsoever--cell phones, pagers, etc. Once I parked my car, I had to go to an unlabeled metal door with a tiny camera on the top. Security guards would buzz me in and require me to sign in at their station. Then I would get buzzed in to the main data center room that contained another room inside of it. From there, I had to enter a password into another security system and place my palm on a palm scanner. Inside this room was another security guard--I would have to sign in with them, too. Then I would enter a different password into another security system, and place my head in front of this retinal scanner. This would buzz me into another room with the cages for each of the clients. There was a padlock on the cage, behind which were our servers. The servers required two separate smart IDs to be placed into an external card reader so that there had to be at least 2 people there to perform any maintenance. The servers themselves were locked down pretty tightly, too. It all seemed pretty insane as far as security goes, but I understood--these computers contained every credit card for the credit card issuer.
Well, after about 3 days of going to this data center, everyone got to know me. They would sign in for me to speed up the process. The security guard behind the door with the palm scanner used to get very hot, so she would often block the door open, thus defeating the palm scanner. The retinal scanner also had problems, often requiring about 3 tries before it would read correctly, so that door was often blocked open, too. Then, one day one of us had forgotten our smart card. We started cursing, as the round trip to pick up the card was about 45 minutes, so we tried it with only one smart card. Bingo. It worked. So then we tried it with no card. Seems the card readers weren't functioning properly. So, overall, we were able to defeat all of the security measures except for the padlock, and all because the security staff (getting paid 2 bucks above minimum wage, no doubt) all "knew" us. In my humble opinion, it would have been far smarter to *not* have the security guard in the foyer behind the palm scanner. After all, social engineering is probably the most common form of circumventing security.
Another funny thing about this was that we had a rather difficult security audit for all code releases. We had a bunch of ex-NSA employees working for us that were rather good about it, too. We would also hire outside auditors to do reviews of major code releases. It was all fantastastic, except for one thing: code patches didn't get the same scrutiny as code releases. In fact, they got none. Well, in order to expedite the release of one particular feature (that required emailing confirmation to customers), we packaged it as a "patch". No security audits. And for something that required the installation of a mail server! Furthermore, the code base had access to the record-level encryption used to store the credit cards. So, basically, if I had wanted, I could have installed a bit of code that would have decrypted all of the credit cards of users of our software and emailed them to a third party. I could not believe it. It's a good thing I have what I consider to be high moral and ethical standards.
I realized through this ordeal that security measures are not put in place to ensure security. They are put in place to give people the perception of security. And, furthermore, automation and removing the human element are good things for security. People should be used to monitor and oversee automated security, not to be actively involved in that automated security.
One word: immigration. The US still has far and away the best universities in the world. There are a lot of reasons for this, but probably the biggest one is that professors can make several times the money in the US as they can in the UK or in mainland Europe. Additionally, there is a great deal more investment capital for startups in the US, encouraging many of the brightest to start companies in the US. As a result, many of the best European researchers come to the US for their research, and stay here to start high-tech companies. This is not a new phenomenon; simply look at some of the most famous US scientists over the years to see how many of them are immigrants (such as Alexander Graham Bell and Albert Einstein).
Last I checked, something like 60% of the graduate students at these US universities are from other countries. A significant number of those stay after achieving their degrees. Even though the primary education system in the US falls short of many (most?) developed countries, the US has succeeded in attracting these foreigners, with many of them immigrating after achieving their degrees.
The biggest threat to this cycle? US foreign policy and immigration policy. With this "global war on terror", the US has made immigration far more difficult. At the same time, the image of the US has suffered tremendously as a result of US foreign policy, trade policy, environmental policy, and human rights policy. If the US stops this cycle of immigration as a result of its self-centered governmental policies, then, yes, the US very well may lose its edge in the global economy.
Yes. After all, it's so difficult from the naming convention to tell that the Solo has one core, and the Duo has two cores. Yes, the naming convention is pretty confusing. With a name like Turion, it's so obvious that it has only one core. (OK, so I'm ignoring the Core 2 Duo name...that one *is* pretty f'ed).
My concept of "fair value" for a corporation is the net present value of their future free cash flow, plus their assets, minus their liabilities.
that's not what the grandparentposter said. He said that market cap is not a good indicator of how impactful a fine is. For example, it might affect existing credits adversely, increasing the cost of existing lines of credit as their credit rating drops. That's an effect unrelated to market cap, and related more to things like the actual assets and cashflow.
So, since the fine affects the credit [rating] adversely, according to you, this affects their cost of credit, and therefore impacts cash flow. Well, lo and behold, my concept of "fair value" says the market capitalization should drop to reflect their cash flow. So, the market cap *is* affected. In fact, my basic argument is that the market capitalization becomes a good proxy for the impact of the fine.
Anything and everything can and will influence the stock quotes themselves, and influence market cap, including whether the weather is nice or not, if some trader is having a good day, etc.
And if the market cap (i.e. share price) gets out of whack because a trader accidently puts in a sell order for 10M shares instead of 10 shares, then this will present an arbitrage opportunity. Arbiteurs look for these events and *will* correct this. Arbitrage does fix this, although the fix isn't instantaneous (as those who believe in perfectly efficient markets would have you believe). How fast it gets fixed is a function of the liquidity of the market. The more liquid the market (i.e. the greater the # of buyers and sellers, and the greater the volume of trading), the faster the correction will occur.
I'll give an example where this simply won't work. Consider the largest owner of commercial real estate (shops) in, say, The Netherlands (Maxeda). Maxeda's assets, its real estate, are worth more combined than its marketcap! Madness, you'd say! The invisible hand of the market should correct this outrage! Either Maxeda is worth more, or a corporate raider should split the company up and sell it in parts.
The value of the corporation is not a function of thier net assets alone (also known as their book value). As I said before, the FCF valuation method shows that it is a funciton of their net assets plus the net present value of their future free cash flows. If a company is trading below book value, then I can fathom a number of reasons this might be the case. First, the accounting value of the assets may be incorrect (this is basically what you are saying regarding Maxeda...splitting the company up & selling it off won't return the accounted value of the assets, because the accounting value is greater than the liquidation value). Second, the company might be hemorrhaging free cash flow. If an arbiteur sees this second case (as Kirk Kerkorian does with GM), the arbiteur might decide to take a large minority interest and change the management so that it won't hemorrhage cash (as Kerkorian is doing with GM), or the arbiteur might decide to split up the company & sell it off. In either case, the goal is to prevent erosion of the asset value that results from negative cash flows.
The same goes for market cap, as the grandparentposter eloquently explained; if you suddenly dump every share on the market, you won't get a good price for them - and it's doubtful the price would recover. Conversely, if you try to do a hostile takeover of a company, expect to pay a premium over the market value, like Mittal recently did to acquire Arcelor. There was no magic arbitrage going on there that made sure Mittal paid the right price; in fact, they probably paid too much, but they expect to make good on that by using their controlling interest to do new and exciting things. And that is merely a gamble.
In the case of Mittal and Arcelor, the Mittal family believes that acquiring Arcelor presents opportunities for new future cash flows that would not be poss
What about arbitrage? Basically, if a stock goes into freefall for no apparent reason other than a change in supply/demand for stock trades, then the value of their shares would be lower/higher than the value of the company. This is a classic case of arbitrage, and will eventually lead to an arbiteur to buy (if it is undervalued) or short (if it is overvalued) the stock. If the company is significantly undervalued, an arbiteur might decide to purchase a significant portion of the outstanding shares and attempt to divide the company into pieces & sell it off. This is what a corporate raider does.
What is a fair value for a company's shares? This is a subject of great debate. However, based on a statistical analysis I did comparing a slew of the more common ways of valuation for a company, I found that the most accurate valuation method is FCF (Free Cash Flow)--basically, count up all of the company's future free cash flow (i.e. sort of like profits, but based on true cash flows, not on accounting profit) discounted for inflation, and subtract the company's net liabilities. A one-time fine will reduce the value of the corporation by the amount of the fine plus any perceived impact the fine will have on future free cash flows (i.e. the fine will likely hurt the image of the company, and therefore reduce future sales), and will impact the proper valuation for the company. So far this morning, since this fine was announced, the price of the stock has dropped $0.358. Multiply this by the number of outstanding shares (10.20 billion), and you get a financial impact of something like $3.65 billion. Of course, this isn't the only thing impacting the company's business (a 20% chance of Vista not shipping also plays a part). But, to say that Microsoft's market capitalization is not affected by this ruling is naive at best, and outright wrong at worst.
Supply and demand do have short term impacts on share price. But, in the long run, the arbiteurs will restore the share price back to normalcy. The more liquid the market, the quicker this process of restoration will occur. Market capitalization is not a joke number, since the owners of a company's stock get to vote to control/guide the corporation.
Since you're giving out your books for free, can I have a copy of each of them?
Oh, man, I think I had better buy some silicon dioxide futures then. Or, I suppose, we could always just go to a desert & pick up some sand...
Kidding aside, manufacturing cost curves for solid state follow the First Corrolary to Moore's Law--that the price per transistor halves every 18 months (Moore's Law states that the # of transisters doubles every 18 months, but the corrolary also proves true). According to this article, the period for photovoltaic cells is about 4-6 years. Hard disk drives seem to be following a period of 2-5 years, with a greater deal of variance than with solid state. It is inevitable that solid state will be cheaper than hard disk drives at some point.
The same holds true for LCD panel monitors. Some day, wall-sized LCD panel monitors will be only slightly more expensive than wallpaper, and we'll be living in a Total Recall world.
Back about 15 years ago, I calculated the crossover point for solid state vs. hard disks as occuring in 2008. I don't think I'll be too far off...maybe 2010?
Great question, especially considering Sun now offers N1 for free. I guess the massive amounts of configuration & app support required for it still make it expensive, and Sun is hoping to get some of that work in the form of a consulting contract.
This question, far more than one about the absolute cost of Cassatt's solution, is the one that will decide whether I choose Sun N1, Cassatt, or whether I choose to build out my data centers so that each app has dedicated hardware. After all, the $5k per server costs would cover probably 1-3 months of the total costs in the real world of running a server; if I can eliminate those costs, $5k/year is a bargain.
Try cooling a dual proc computer without the case sounding like a jet engine...better to go with dual core. At least for the type of procs that can decode HDTV
$2276 is for a dual HDTV tuner setup. The case is so that it looks pretty and doesn't sound like a jet engine. The $300+ cpu gets them a dual core cpu (see below).
I'll agree on the RAM, although if the computer is used solely for a media center PC, 2GB of memory isn't that helpful. And I'll also agree on the hard drive, having just purchased a $200 400GB hard drive myself for my media box.
Honestly, given the proper motherboard, onboard sound with digital outputs going to the DAC in my stereo, and I see no need for a $100 sound card.
Just try to decode an HDTV signal on a $150 CPU. It can be done, but not if the PC is doing anything else. They went dual core, I would bet, so that they can decode an HDTV signal while simultaneously recording 2 HDTV signals and a regular signal, all while doing pretty much anything else without skipping a beat.
It's not perfect, but it is fairly reasonable. It does feel like they picked the components of their advertisers, though.
I'm afraid you are 100% incorrect.
.95c, then turns on a flashlight pointed in the direction of travel, a casual observer in front of them would see the light travelling at 1.0c, and the "spaceship" travelling at .95c. To someone on the spaceship, the light will be leaving the spaceship at 1.0c.
.05c. The wavelength of light will appear different for each observer, due to the differing apparent distances and timeframes for the two observers.
If a "spaceship" is travelling at, say,
How is that possible? Time/space dialation. In effect, time appears to slow down for an observer on the spaceship (relative to the person on the ground). Additionally, space appears to lengthen for the observer on the spaceship, so that the distance between the spaceship and the "fixed" appears longer. The net effect? The observer sees the light leave the spacecraft at 1.0c, and the fixed observer sees the light separate from the spaceship at
Actually, Moore's Law relates to the number of transistors per unit area, not processing power or clock speed. The corrolary to Moore's Law indicates that the price per transistor halves every 18 months. Nothing is ever said about clock speed or processing power. Check this link for the actual wording. By the way, Moore's Law still seems to be holding, and with the proliferation of multi-core computing and whatnot, I expect Moore's Law will hold for a while to come. The difference now is that we will start to make better use of multiple layers on a single piece of silicon, instead of working towards smaller transistors. We are, indeed, starting to bump up against the laws of physics with the size of transistors, but we're only starting to explore the benefits of creative packaging.
Why don't you look at a longer duration than 1 day?
This is not a case of air not being linear. This is a case of the summation of wavelets resulting in beaming, as opposed to a radial dispersion pattern. This is true of any wave (electromagnetic, water, or air/gas). If the width of the transducer is less than 1/2 wavelength, you get a radial dispersion pattern. Likewise, if the tranducer is greater than 1/2 wavelength, you get no dispersion (hence beaming). This is the reason why speakers have enclosures. The enclosure prevents the backwave from cancelling with the front wave. This is also the reason why eclosureless speakers (like those from Martin Logan and Magnepan) tend to be fairly large.
The combination of frequencies is known as aliasing. Aliasing is what allows sonic frequences to be created from two supersonic frequencies.
There's nothing about linear versus nonlinear behavior of air in effect there as far as I know.
One advantage of this effect, for audiophiles (at least once the patents expire--I believe they were awarded in the mid-90s), is that a single tranducer can reproduce then entire audio spectrum. 20Hz-20kHz is 8 octaves, a mammoth feat for any single tranducer to reproduce in linear fashion (by linear, here, I mean that doubling the input power doubles the output power). However, move that from 100kHz-120kHz (resulting in the same 20-20K range when using this heterodyne concept), and the tranducer only needs to create roughly 1/4 of an octave linearly in order to recreate the entire audio spectrum. That task is far easier to do than doing it the "old fashioned" way.
In about 10 years, I would expect to see a plethora of audiophile-quality speakers that consist of an incredibly tiny flat square panel tranducer, coupled with a reflecting device that reflects the sound in a radial pattern. Or perhaps we'll end up with directional reflectors that are able to follow us, allowing a "Minority Report"-type directional advertisements.
What exactly does heterodyning have to do with air molecules behaving nonlinearly? I understand acoustics fairly well, and understand that if the baffle around the transducer is greater than 1/2 wavelength, you get beaming instead of radial dispersion. This, combined with aliasing effects, is what I understood caused heterodyning. I've never heard anything about air molecules behaving nonlinearly as a cause. Can you elaborate?
And you weren't using the concurrent tenured GC algorithm. Here's a decent article covering JDK v1.4's garbage collection algorithms.
Your apparent description of the problem (heavy swapping, a GC that runs at low priority) is simply wrong. The problem you were seeing is that there was no available space in the "new" heap space for new objects, so objects were immediately promoted to tenured space. Once the tenured space gets to a certain point (I think the default is 75% full), the tenured GC runs and stops the JVM cold in its tracks (if a naive GC algorithm is used). If there is no space to be reclaimed during a tenured GC, it can take a really long time to run. If your heap is large (>512MB, in my experience), this can take >10 seconds.
Use the "concurrent" GC algorithm, and these "stop the world" delays are reduced to less than 0.2 seconds, in my experience.
To use the concurrent GC algorithm, start the JVM with this command-line option: "-XX:+UseConMarkSweepGC". Also, you probably want to do some analysis of the JVM churn of memory. There is a good chance your application may require more "new" space, or different ratios between new & old spaces, and for the survivor space.
If you are having problems with swapping, buy more memory. I don't think porting the code to C++ will help.
A better solution to your problem is to quit doing XSL transforms, XPath, or DOM processing of XML documents in a server-side application. The app where you saw these delays did heavy XML processing, right? Yeah, I thought so. Try using a SAX parser, and re-implement your transforms to work on small blocks rather than entire documents. The problem is you are creating too many new objects (something that XML processing does a lot of) without freeing them, causing your "young" space to get filled up before the "young" GC can run. This causes the tenured space to get cluttered with "new" objects, resulting in fragmentation and increased frequency of tenured GCs. That, in turn, causes a "stop the world" response that compounds the problem, since the JVM is spending >30% of its time attempting tenured GCs when there is nothing it can garbage collect (since it is filled with "new" references).
In a word, server-side XML processing is "Bad".
For most non-American speakers of English (as well as many Americans), the Oxford English Dictionary is the reference for the spelling of English words. Well, the OED lists the -ize spellings as the primary spellings for most words (but not analyze/analyse). Here is a link for you.
On a related note, the word 'soccer' was introduced by the British, as well. It seems that the move towards "-ise" and football are attempts by the Brits to appear more like the French.
Sounds like it's not a far cry from "500 companies we think are cool".
This approach of measuring corporate valuations is called FCF (Free Cash Flows), and should be based upon free cash flows, not on dividends. It is very commonly used in the investment community, and shows a *much* higher confidence factor when compared to actual valuations than either trailing or leading P/E valuations.
Looking at http://finance.yahoo.com/q/ae?s=GOOG we can see that Google currently has a trailing P/E of about 121, but a forward P/E of around 46. That forward P/E really is not too outrageous for a company in a high-growth phase.
Take a look at the FCF calculations and you can see that Google may even be undervalued, depending on your estimates of future growth and profitability. Google has shown fairly consistent gross profit margins, and improving operating profit margins as they have grown. Furthermore, they are showing a growth rate of >40% per year in revenue. If they can maintain that for 3 years, then reach a steady state in growth & profit margins, their $300+ stock price is actually fairly valued. If they can exceed that growth, sustain the growth for longer than 3 years, or improve their profit margins, then an even higher stock price is warranted. Remember, stock valuations aren't based upon what the company did in the past, but what investors think will happen in the future.
Given that, I think everyone here has missed the boat. Their search engine is wonderful, and drives a lot of their business. But remember that ad sales from their search engine accounts for only about 1/2 of their revenue. The other 1/2 comes from Ad-sense selling to other websites. If Microsoft releases a magical search engine that manages to steal 100% of Google's search engine business, Google still would only lose 1/2 of their revenue. Based on their profit margins, I would bet they would be able to remain profitable in the event of such a change, too.
Google is an advertising company. They make money by driving traffic to their (free) web sites, and by selling their superior ad technology to other websites. For all that everyone may hold Google's search engine technology in high esteem, their targeted advertising is equally impressive. Combine this with the multitude of data streams they have to collect information about the visitors to theirs and other websites that use AdSense, and they have a killer product that cannot be matched easily by any of their competitors.
$300 is a fair valuation. Anyone using P/E ratios to demonstrate that a company is either over- or under-valued is, imho, a moron. P/E ratios don't capture growth rates, and don't reflect the amount of free cash flow a company can generate. In the end, it is free cash flow (and not profits) that drive a company's worth. It is very difficult to "cook the books" with respect to cash flows, also, so FCF valuations will be more immune to Enron-esque bookkeeping, too. Don't believe me? Look at the FCF generated by Enron in 1996-2000. The writing was on the wall then. And, the writing is on the wall now with Google. Google is a sound company with a killer business plan. Furthermore, they have an excellent record at execution that I believe indicates they will be able to sustain their growth for a long time to come.
Damn, no mod points. I didn't laugh at the original (grandparent) comment, but yours had me in stiches. Who says you can't convey sarcasm in written form?
So if you are a haliburton employee and you are against the war in iraq you should be fired?
If you throw in speaking about it in public (and I think the grandparent post implied that), then you've got a situation where a person is "publicly stating an opinion that is contrary to the business interests of the corporation". Is that acceptable to fire someone that voices such an opinion? I don't think so. I think the grandparent was right on target with their statement.
However, if the person was working in a right-to-work state in the U.S., then I believe the employer has the right to fire that person for any reason so long as it is not discriminitory. I don't believe this would count as discriminitory. I believe this argument was used as the justification for the private corporation that fired an employee during the U.S. presidential election primaries because the person attended a Bush rally wearing anti-Bush shirts.
Man, you are a Java developer and you've never heard of Eclipse? Where have you been hiding? Let me guess--you've never heard of SWT, either?
If you are a Java developer and you haven't heard of eclipse, download it. It's the open-source IDE originally from IBM. It seems to be the IDE with the most traction in the Java world. IIRC, even Borland and BEA have decided the next versions of their respective IDEs (JBuilder and Workspace) will be simple plugins to Eclipse.
It's quite a nice IDE. Really. There's a ton of useful plugins for it, too. I'm not so sure it's any better, technically, than NetBeans, but it sure has a critical mass of 3rd party support.
Yes, the Jetty lead works for Gluecode now, I think. That's one of the reasons the JBoss/Jetty integration has been dropped. There's bad blood between JBoss & Geronimo.