Analysts rarely issue sell ratings since they do not want to say negative things about companies that their firm might do business with in the future. So normally, you should read an analyst rating as one step below what they actually say: hold (their lowest rating) = sell, accumulate = hold, buy = buy (maybe).
Second, there is the fact that analysts almost never say "sell." According to Thomson Financial/First Call, fewer than 2% of all analyst recommendations are "sell" or "strong sell." This problem can be partly solved through translation: Even minimally savvy investors sense that "strong buy" means "buy" and "buy" means "maybe you should buy" and "hold" means "sell." When an analyst does say "sell," he means something like "You idiot, you should have sold that stock six months ago."
--Pat / zippy@cs.brandeis.edu
Re:SGI is still in business?
on
New SGI Altix 3000
·
· Score: 5, Interesting
Thanks for the suggestion. So I did take a look, and the answer to the question "how are they staying afloat?" is "they aren't. They are steadily hemmorhaging money."
I posted my question because, as early as 1996 or 1997, it was clear that commodity machines were going to kill SGI. I'm just amazed that they're still alive.
Here's their income statement, balance sheet, and cash flow statement. As of today, For three out of the previous four quarters, they had sales growth (sic) of -20% or worse. The three analysts who cover this stock have a hold rating, which in analyst-speak means sell.
--Pat / zippy@cs.brandeis.edu
SGI is still in business?
on
New SGI Altix 3000
·
· Score: 5, Interesting
What is keeping SGI afloat? Service contracts on existing machines?
I agree with every point you've made. In 1996 I started to believe that commodity hardware would kill SGI. Today, with even SGI's core customers doing rendering on linux clusters, I really can't imagine who is still buying SGIs.
I'm amazed they haven't gone bankrupt yet. Perhaps they have a big revenue stream from servicing their existing, dwindling base?
You can get high accuracy with a small representative sample. You can adjust for the bias in a skewed (non-representative) sample. Check out any intro text on statistics.
I used to manage the logs for a company that collected web usage on 150k users.
You can normalize for your sample by comparing your group against a known baseline. What is interesting to me is that the "normalized" groups like what Neilsen uses are tiny (3000 users) and thus have their own flaws (undercounting Goatse visits since their population of users is so tiny they're missing out on all sorts of behavior).
Here are some things you could do.
1) track e-trade, schwab, and yahoo finance to see what stocks people are checking - a good way to anticipate market changes/volatility
2) monitor site traffic for e-commerce sites to gauge how much business they are doing.
3) accurately measure how many click thoughs banner sites get (almost none)
I wouldn't be surprised if the Bank of Brazil is switching because Brazil's banks desparately need to save money (Brazil and its banks are going through a huge financial crisis).
I worked on some projects with the Internet Archive from 1998 - 2000.
The Archive's first storage device (circa 1996) was a large StorageTek tape robot with a multi-gigabyte disk cache to handle user requests for archived pages. As drives and processors became cheaper, it became more interesting to use them instead of tape. The cost penalty of using drives over tape is only 2x - 3x, with the enormous win of increased bandwidth and decreased latency (when the request queue for the bot got large, the wait time for a page could be 16 hours. With disk, it's a fraction of a second).
The first hard-drive based Archive storage used multiple 4U and 5U 12-20 drive Linux/FreeBSD boxes with ~80G IDE drives and Promise cards.
Drive density is greater now - you can get 200G IDE drives and 320G IDEs are on the way, so you can use regular PCs as opposed to custom or niche-market (rackable server) boxes.
If the military satellites are at all like commercial imaging satellites, you wouldn't be able to tell which photos are still classified.
Commercial spy satellites like Quickbird don't take pictures of every square of earth. Instead they photograph sites users request photographs of.
And they might photograph the same site at different points in time.
So unless you noticed something like that an important Soviet base was completely missing from the set, you wouldn't be able to tell what images were still classified.
I thought reel to reel tape had a lifespan of under a decade until serious data loss.
If the tape is already 30 years rotted, and was initially thoroughly wiped, is there much of a chance that they can recover anything?
--Pat / zippy@cs.brandeis.edu
When the original Myst came out, I thought about how it would be a great evil project to make a frag-fest map based on the island once realtime 3D graphics caught up with 1995-era rendered images.
Ignoring whether RealNames spent money well, had a good idea, or took the right engineering approach,
they were still screwed.
Why? Because their approach required a tie-in with the browser.
There are only three browsers in wide use: IE, AOL, and Netscape/Mozilla.
IE is the overwhelming favorite, to the extent that many businesses completely ignore all other browsers. If a company can serve 90% of the users with one platform, then the company is likely to stop right there as the remaining 10% of users/platforms require just as much effort as the first 90%.
In RealNames' case, it's not just about their deal with Microsoft, it's that they had no one else to do business with. Netscape and others are not viable competitors to IE. Without serious browser competition, IE doesn't have to worry about someone else having a feature that they they need to match. In short, RealNames had to deal with MS so they could serve the 90% of users on IE.
Yes, we all know MS eats competitors and partners alike. And when MS is the only browser worth supporting, who else are you going to deal with?
Most users think of the default home page as "the internet" because that's what they see when they log on.
It honestly doesn't occur to them that they could point it to anything else. It's what they know, and as far as they know, it's what they have to go to in order to connect to the net.
Many people have flamed RealNames for working semi-exclusively with Microsoft, but really, they had no choice. They had to depend on Microsoft because Microsoft controls the browser market!
Who else makes a widely used browser? Netscape?!? Sure, everyone on Slashdot uses Mozilla, Conqueror, iCab, and lynx, but...
Try Tools -> Related Links. This points to Alexa Internet. Sure, it's not as high level as the browser home page button, but it is a service where MS has for years used an external provider.
Slashdot Mirror
This is a well-known practice. Here is what Forbes has to say about it:
--Pat / zippy@cs.brandeis.edu
I posted my question because, as early as 1996 or 1997, it was clear that commodity machines were going to kill SGI. I'm just amazed that they're still alive.
Here's their income statement, balance sheet, and cash flow statement. As of today, For three out of the previous four quarters, they had sales growth (sic) of -20% or worse. The three analysts who cover this stock have a hold rating, which in analyst-speak means sell.
--Pat / zippy@cs.brandeis.edu
What is keeping SGI afloat? Service contracts on existing machines?
--Pat / zippy@cs.brandeis.edu
I second this rant.
I always thought it should be called Moore's Observation.
--Pat / zippy@cs.brandeis.edu
+1 Insightful
Here are the things that suck about Sprint (and other providers who do similar stunts)
1) billing is rounded up to the nearest minute. 4 sec call? 1 minute.
2) prime time (non-night) hours: 7am to 9pm
3) spotty digital coverage
4) analog = roam = large per-minute fees
--Pat / zippy@cs.brandeis.edu
You've certainly asked the right group.
--Pat / zippy@cs.brandeis.edu
I agree with every point you've made. In 1996 I started to believe that commodity hardware would kill SGI. Today, with even SGI's core customers doing rendering on linux clusters, I really can't imagine who is still buying SGIs.
I'm amazed they haven't gone bankrupt yet. Perhaps they have a big revenue stream from servicing their existing, dwindling base?
--Pat / zippy@cs.brandeis.edu
--Pat / zippy@cs.brandeis.edu
You can get high accuracy with a small representative sample. You can adjust for the bias in a skewed (non-representative) sample. Check out any intro text on statistics.
--Pat / zippy@cs.brandeis.edu
I used to manage the logs for a company that collected web usage on 150k users.
You can normalize for your sample by comparing your group against a known baseline. What is interesting to me is that the "normalized" groups like what Neilsen uses are tiny (3000 users) and thus have their own flaws (undercounting Goatse visits since their population of users is so tiny they're missing out on all sorts of behavior).
Here are some things you could do.
1) track e-trade, schwab, and yahoo finance to see what stocks people are checking - a good way to anticipate market changes/volatility
2) monitor site traffic for e-commerce sites to gauge how much business they are doing.
3) accurately measure how many click thoughs banner sites get (almost none)
etc.
--Pat / zippy@cs.brandeis.edu
I wouldn't be surprised if the Bank of Brazil is switching because Brazil's banks desparately need to save money (Brazil and its banks are going through a huge financial crisis).
--Pat / zippy@cs.brandeis.edu
Brewster did the router hardware for the Thinking Machines CM-2. I think Tamiko Thiel did the case design for the CM-2 and the CM-5.
http://mission.base.com/tamiko/cm/index.html
--Pat / zippy@cs.brandeis.edu
I worked on some projects with the Internet Archive from 1998 - 2000.
The Archive's first storage device (circa 1996) was a large StorageTek tape robot with a multi-gigabyte disk cache to handle user requests for archived pages. As drives and processors became cheaper, it became more interesting to use them instead of tape. The cost penalty of using drives over tape is only 2x - 3x, with the enormous win of increased bandwidth and decreased latency (when the request queue for the bot got large, the wait time for a page could be 16 hours. With disk, it's a fraction of a second).
The first hard-drive based Archive storage used multiple 4U and 5U 12-20 drive Linux/FreeBSD boxes with ~80G IDE drives and Promise cards.
Drive density is greater now - you can get 200G IDE drives and 320G IDEs are on the way, so you can use regular PCs as opposed to custom or niche-market (rackable server) boxes.
--Pat / zippy@cs.brandeis.edu
Commercial spy satellites like Quickbird don't take pictures of every square of earth. Instead they photograph sites users request photographs of.
And they might photograph the same site at different points in time.
So unless you noticed something like that an important Soviet base was completely missing from the set, you wouldn't be able to tell what images were still classified.
--Pat / zippy@cs.brandeis.edu
Gee, hope those tanks aren't near empty when you fire.
--Pat / zippy@cs.brandeis.edu
--Pat / zippy@cs.brandeis.edu
I thought reel to reel tape had a lifespan of under a decade until serious data loss. If the tape is already 30 years rotted, and was initially thoroughly wiped, is there much of a chance that they can recover anything? --Pat / zippy@cs.brandeis.edu
We're there now.
I see a great need.
Just imagine detpacking all those pesky puzzles.
--Pat / zippy@cs.brandeis.edu
Why? Because their approach required a tie-in with the browser.
There are only three browsers in wide use: IE, AOL, and Netscape/Mozilla.
IE is the overwhelming favorite, to the extent that many businesses completely ignore all other browsers. If a company can serve 90% of the users with one platform, then the company is likely to stop right there as the remaining 10% of users/platforms require just as much effort as the first 90%.
In RealNames' case, it's not just about their deal with Microsoft, it's that they had no one else to do business with. Netscape and others are not viable competitors to IE. Without serious browser competition, IE doesn't have to worry about someone else having a feature that they they need to match. In short, RealNames had to deal with MS so they could serve the 90% of users on IE.
Yes, we all know MS eats competitors and partners alike. And when MS is the only browser worth supporting, who else are you going to deal with?
--Pat / zippy@cs.brandeis.edu
It honestly doesn't occur to them that they could point it to anything else. It's what they know, and as far as they know, it's what they have to go to in order to connect to the net.
--Pat / zippy@cs.brandeis.edu
Who else makes a widely used browser? Netscape?!? Sure, everyone on Slashdot uses Mozilla, Conqueror, iCab, and lynx, but ...
--Pat
--Pat / zippy@cs.brandeis.edu
--Pat / zippy@cs.brandeis.edu
For instance, here's a list of intentionally open wireless access points around SF
--Pat / zippy@cs.brandeis.edu
How many viable browser and search companies are there to partner with?
1) Microsoft/IE
2) Google
3) AOL???
There just aren't many vendors to make deals with here.
--Pat / zippy@cs.brandeis.edu