4) Here's a biggie. Science costs a LOT of money--the only two groups that can afford it are governments, and expensive biotech companies.
[snip]
The most reasonable way to read this comment is
that Big Science costs Big Money. And that is
basically true. And there is no doubt that certain truly Huge projects, like the HGP, cost
a lot of money. What is quite interesting about
the HGP is the fact that there was surprisingly little disagreement that the project was worth doing, and that a public consortium should do it.
And this was yet another big reason why Celera annoyed so many people.
But even though the HGP was as Big a science as
it gets, the thing that really takes your breath
away is how Small some of the
most crucial ideas were; ideas that ending up speeding the whole enterprise by years. This, for me, is one of things that makes Big Science worthwhile: there is
so much Small science waiting to get out. Getting
any grant to do anything approved these days is an exercise in masochism. It's much easier to burrow
into a nice, warm, big project than do the small
stuff by yourself. I think this is why much of the really big stuff ends up getting done either
by accident, or by somebody willfully doing something other than what they were supposedly getting paid for.
Of course, Big Science (whether public or coprorate) doesn't always end up appreciating the
nuggets of innovation that pan out from time to
time. Shotgunning the genome was something that the HGP didn't really think was going to fly, or
at least fly so fast, so when push came to shove,
Ventner walked out the door.
Google's interface for web searches is _useless_ for usenet.
Couldn't they keep the existing Deja functionality until they had something decent to offer?
I'll actually hazard a guess here that the answer is "no". If we're talking about Deja, it looks to
me like that plane was going down so fast (with
both engines burning, noxious gasses filling the
cockpit, etc.) that the cost-effectiveness of keeping up their web presence (which was not *that* cheap) was about zero. They knew they were
going to assimilate the database, that the re-structuring required was massive, that their own current archive was possibly in better shape than the one belonging to a not-really-going-anymore concern, and that there
was no point in waiting to kill off something they
had no interest in supporting in the future.
I can't believe how completely un-sympathetic to the needs of existing Deja users this sudden, and obviously not-at-all-thought-out, gutting of Deja is on the part of Google.
I'll grant you that this sucks a bit, but I'll be
willing to bet at least a little bit that the rapid crash-and-burn had at least as much to do
with deja's deteriorating situation as it did with
Google's arrogance/lack-of-planning/whatever. The
Google press release on the Deja
acquisition basically states that they acquired all of the "significant assets" of deja.com, which is a fancy way to say that this was not a merger, but at best a firesale, and that
the insignificant assets are probably stuff like
your former deja account.:-(
[...] what do you think the odds of Google acquiring such data are?
Gah. I'm not the type to flame somebody for their grammar, but good god... What kind of sentence is this? What you thinking were?
This is a question for...PSYCHOLINGUIST MAN!
To be completely serious, this is a perfectly
grammatical sentence. Indeed, I think it would make my Good
Buddy Robert Kluender beam with joy. Now,
is this kind of thing a piece of cake to parse?
No way: it has what we experts call an unbounded wh-dependency. Indeed, our willingness to torture^H^H^H^H^H^H^Htest
undergraduates with stuff like this is why we make the big bucks.
Now, to prove to you that this sentence is legit,
consider the following:
The odds of Google acquiring such data are small.
I think the odds of Google acquiring such data are small.
Do you think the odds of Google acquiring such data are small or large?
What do you think the odds of Google acquiring such data are?
Does this help any? Now, the real interesting
question is why people would tend to say (4) above
as What do you think the odds are of Google acquiring such data? But I have office hours
in five minutes, so that question will have to wait for another day.;-)
To make this just remotely related to the topic of
search engines and Usenet, I'll point out that long distance dependencies like this one are the kind of thing that can make it infuriatingly difficult to use easy cues like "lack of proximity" to decide that two search terms are truly unrelated to each other. Unfortunately,
solving this one requires you to parse natural
language as it is used on Usenet, which is truly
a frightening thought.
This is a case where we don't really have to
wonder what the free market would do if privacy
rates were publicly traded, since they already
are, in the most high-tech place many of us ever
visit on a regular basis.
In the grocery store. The people who were the big push in bringing you Universal Product Codes, instant coupons based on what you bought, and, let's face it, the entire shopping cart metaphor
on which the internet is based (:-)), have also
brought you, in many markets, the Value Card. (OK, sometimes it's a key-chain or dongle these
days.) The only thing you need to do to get one
is to fill out the handy form...and then, from then on, everything you buy from there is recorded as having been bought by you. Everything from
avacados to zesta saltines, and from birth control
pills to Prozac.
It's a marketer's dream.
Now, you do, of course, get something for
your trouble: sales and price breaks that, in my
experience, never amount to anything more than you could have gotten from the same grocery store before they had the "Value" program, or more than
you could get from a competing grocery store that did not have a "Rewards" program (or whatever they might call it). So let's calculate the value of
your privacy, as estimated by the market:
Hmm, that looks like about Zero, to me. Now maybe
it's possible that when all stores have these programs, and many people are essentially forced to join one to get the kind of prices they used
to get without selling all of their privacy, it's
possible then that we would see market competition
between grocery stores concerning what you get for
your privacy rights. Or not. What we might more
likely see is a vast expansion of the "instant coupons" idea, where your price and my price for
the same good can vary depending on the rest of
our buying habits, and which brands are willing to
shell out a little bit more in price breaks for their least loyal customers.
Yeah, yeah. Now, ask yourself -- why did they want to deregulate in the first place? What was the incentive for deregulation?
Well, because some people thought that the market would become more effecient, and power would become cheaper.
We lived in California for just over 7 years, and
during the period when they finally got the utility de-regulation plan enacted. And, I swear,
I could never figure out why in the world most people wanted to do this. The utility situation
in California (and you could include water in this as well) is a truly classic example of where markets will have a rougher time, because virtually every major change or transaction involves an external party who would not be taking part in the transaction if it were a conventional market.
The building and siting of power plants,
for example, is a problem that comes up almost everywhere, but the costs to other parties of putting them wherever the grid thinks them most efficient are vastly greater in California than in most places. California generates a huge amount of it gross state product from the fact that it *is* California, the idyllic (-looking) paradise. In California, a huge determinant of
the value of any piece of land is, to be quite
frank, the view, and the clarity of the air and water, which are both kinds of "rights" that
are extremely hard to deal with if the holders
aren't a party to the transaction.
Now, what actually happened was the way people became party to these kinds of transactions was through the political system. Most specifically, the PUC and their right to regulate
the placement and operation of power plants and
the prices that could be charged. In return for
this power, utilities were essentially granted a guaranteed rate of return (like most other places), which is a boring but perfectly profitable way to do business. Yes, there is no
doubt that there were inefficiencies in that system, but, because it was political, all the affected parties were involved, and everybody could play. This was the reason why I find it hard to consider the PUC to be truly a central planning agency: everybody could and did put tremendous amounts of pressure
on the PUC to have things their way, but always
got compromise solutions. The problem, of course,
was that some users believed that they could get
better prices through a different system, and that
it was worth the huge monetary cost of pushing hard on the political system in the usual fashion.
This would have been okay, except that the compromise solution that was reached had some spectacular bugs in it that weren't fully appreciated at the time, but which could be (and have been) exploited to the hilt by power suppliers, who were (by law) completely separated from the utility companies themselves.
So, when I see what happened, I don't think of it
as a market failure or a central planning failure,
but as a political failure of the type that so very often crops up in California. Why California more than most other places? I think it is because the stakes are so much higher there than other places, and because there are incredibly strong regional political differences that make the place basically ungovernable. I cannot possibly imagine something like this happening in, say, Iowa. Or in a California that consisted of
two or even three independent states.
Saving state on exit is a good idea, but that can already be done. You may have already seen it - it's the 'document changed; save?' dialog box.
I respectfully disagree completely.:-)
Seriously, truly persistent document storage with automatic infinite undo, the ability to play with any version you want, automatic indexing and a built-in relational view of all your other documents...I'd be in heaven. The true value of persistent state isn't just saving a few keystrokes here and there. The big win isn't in the data; it's in the metadata, the linkage, the
automatic indexing and all the stuff that is truly tedious to do yourself.
In the olden days, this didn't seem practical, but
next year I can probably get a 200 gig hard disk
for under $300. Which probably only means more empty disk space unless somebody figures out that it's okay to blow a meg of storage on
every text file I ever create if that's what it takes to provide all the real services people want in
a file system (or document system). The capacity of that 200 gig disk is convincingly larger than the entire unindexed capacity (in bits) of the human memory that will serve you so conveniently for decades, and may (finally) be big enough to
hold the index, too.
Now, lest any of you believe I'm condoning some kind of device for idiot, keep in mind that I've
said nothing about the beautiful query language
I have in mind. The use of that is what will
separate the experts from the novices...
In contrast, that 24% profit for Microsoft is after all salaries and expenses. (Given the marginal costs associated with their product, it's damn near 100% using the reasoning you're employing.) It's no wonder that, until recently, they were considered one of the seven wonders of the economic world.
Like, wow; something finally sunk in for me.
As far as what aspect of Microsoft is a wonder
of the economic world, I would have though the
answer was: that absurd P/E ratio. It's now
down to 33 (or up to 33 if you only consider the
last few months). If Microsoft, with complete
market domination, can only grow as fast as the
PC industry, then their P/E ratio really shouldn't
be greater than the PC industry (check), and the
PC industry, now that we're nearing saturation, can't grow much faster than the economy at large,
which is worth a P/E of like 22 these days. So where is the extra growth going to come from? If
their margins are really 24%, it's not likely to be from an increase at the margin, and it's not coming from an increase in market share, and it's not coming from an increase in market size. Looks
pretty bad to me. So if we were to reduce their
margin to like 6% (nothing to sneeze at), hold their P/E to be equal to the market average, and
put the market P/E back at 15 where it has tended
to be, I get a target price for MSFT at...
About $7.50 per share.
I think that's a bit low,
but it get me rather worried about what's going to hit the fan when and if the market really comes back down to earth.
Now you have a changing of the guard at the DofJ, isn't it possible that they initiate settlement negotiations, this time much more favorable to MS?
Can you do that? I mean, normally if you've prosecuted a case and won you don't go out and offer to settle unless you are pretty sure the appeal will overturn the judgement, right?
This has been mentioned as a legal possibility,
but it has the following important limitation:
Any settlement has to be approved by
the trial court judge. It wouldn't
be inconceivable that Judge Jackson would approve
a settlement that would end the Appeals process (remember: he's a conservative Republican and a Reagan appointee), but given the contempt for his court that he detected in Microsoft's conduct of
the original case, I wouldn't bet on him giving up too much without a pretty thorough confession of wrong-doing by the "we did nothing illegal" MS establishment.
[After lots of detail on how his physics book typeset with jury-rigged tools went down in flames, especially on the web...]
As far as LaTeX,
it's never going to be learned by more than 0.01% of the world's population,
it represents a 1970's-style approach to making a user interface (ooh, you mean I get my own terminal instead of having to hand someone a stack of punched cards?), and
its aggressive stance on separating form from content means that you have to jump through hoops to make a complicated layout turn out how you want it.
You give me the giggles.:-)
Seriously, you sound very confused to me. Your
hand-baked approach to putting physics on the web didn't work so well, because, as you point out, there was no way to separate form from content that way. But then you slam on TeX for, well, trying to enforce that useful separation.
The second objection is (I think) to the way that
TeX actually dares to compile your document, report errors, and work essentially in batch mode.
Yes, that can be infuriating at times, but the advantage is that, at the end of the day, your work really could be device independent. Giving up that is giving up the farm, in my opinion.
The first objection is the one that I found most amusing, though. Like, so what if only one person in ten thousand ever learns to use LaTeX? Many fewer people than that ever create anything that needs it. Heck, only a few people in a hundred ever create anything much at all, and it's a given that fewer people know how to use tools than know how to appreciate the results of tool use.
We are the geeks; the people who use tools.
The people who use tools to make tools.
The people who use tools to design tools to make tools. We are the lords of the Shell, the
emperors of Perl, we dominate the DOM.
You're totally right that bitmaps aren't the way
to do equations (or lots of other things). I can
agree that LaTeX isn't perfect or even that close.
But it was the tool that showed that not everything had to be set by hand, that math could be free, that something like MathML would eventually become a worthy successor. Worthy, that is, if we're clever enough to design a useful
input device for the stuff...
Name 3 truly useful apps that require X and don't have equally useful alternatives that don't.
This part is easy. With X, you can run apps remotely and display them where you are. So, for
me, anything I want to run on a Linux box at work
but view at home is such an app. Now, I expect that's not the answer you were expecting, but if X doesn't immediately suggest to you the possibility of remote apps, you're missing the point.
Got 'em? Took you awhile, didn't it?
Nope; that part was easy.
OK, now describe what sort of person will be wanting those applications and will also want a Mac in place of a more powerful and much cheaper PC.
Well, I'm married with two kids. My wife vastly prefers the Mac interface, and an iMac was decent enough looking and quiet enough to put out in a public place, saving additional bucks for one of those "hide the ugly PC" computer
thingies. Now, I do kick myself for not waiting another 6 months and getting the iMac I got for
$500 less, but that's the PC life...
I'll bet that was even harder, but you still came up with something, right?
No, that was still pretty easy.
Well then, now the hardest part: Tell me, with a straight face, that there are more than a handful
of these people in the world.
Now the funny part of this is that even though I live in Columbia, Missouri, a metro area with a population of 130,000 or so, I already know a handful of people like this. But so what? Why
should I care how many other people use computers
like I do? In the bad old days, that might matter because software was expensive and and closed source; not these days. I can have a Mac, a BSD
box, a wireless network set-up, a DVD player, and
all the comforts of a quiet-to-silent PC for a price I was more than willing to pay.
I'd like to be free to live in a country where I can pay for health care that doesn't leave me sitting in an emergency room for three hours waiting for a Keflex prescription for strepped throat,
Me too. But, not being nearly wealthy enough to afford so-called "point of service (POS)" insurance, we make do with a so-called "Health Maintenance Organization (HMO)" which has all the
charms of the Canadian system without some of the
advantages.
while homeless heroin-addicts with needles broken off in their arms come in after me, sit beside me, play show-and-tell with their pus, and then get served before I do, despite the fact that I'm a tax payer and they're not.
Maybe it's just me, but isn't it a good thing to get the pus-o-matic heroin addicts out of the public waiting area as quickly
as possible? I mean, even an HMO will do that. Take my five year old son; no, he's not a disgusting heroin addict, but he did today have a glorious case of chicken pox, so the "pus-filled" part applies. Anyway, the HMO had him in and out
of that room in under 5 minutes. But next time
I have to go in there, I think I might try the
old Canadian "heroin needle in the arm trick" to
get faster service...
I'd like to be free to live in a land where what is played on TV and radio stations is based on market demands, not on CRTC 40% Canadian Content regulations,
Well, down here it seems like the market demands a 40% Backstreet Boys Content (BBC); sometimes it's unpleasant to get what you ask for.
forcing broadcasters to play the same really lame Tragically Hip songs
I would feel your pain, but, hey, you also get
to hear Crash Test Dummies! And, uh, Rush? And
Gordon Lightfoot and...uh, you mean to tell me that the music up there looks just like the godawful Air Canada in-flight music program? If so, I think you could challenge this as a human rights abuse, right?:-)
and poorly lit Canadian TV shows over and over again.
OK, now that you mention it, what *is* up with that? Are courses in lighting design just not
offered north of the border (or in many European countries)? Does this have something insidious to do with the metric system? (Nope; Japanese TV lighting is superb.)
Most of all, I'd like to be free to go outside without fearing for my life for 5 months of the year. I don't define quality of living by habitating in a place where you can die simply from going outside without a jacket on.
Then don't come to Missouri; here you have the
Canadian winters plus the special treat of those
dew-points-in-the-high-70s days that we call "summer". We cheat cheat death 8 months out of the year by daring to leave our climate-controlled subdivision houses and SUVs.
...but I suspect that this is merely to give them carte blanche to "kick ban" harrassing folk. Same with their policy of port scans and the like (as they probably don't want to have people contacting their sysadmins saying that there are script kiddies trying to break into their networks).
Yup. This particular TOS is pretty much designed to make sure that they can enforce something at
some time rather than everything at every time.
Specifically, they mention that:
10.4 Sprint's failure to enforce strict
performance of any provision of this Agreement
will not be construed as a waiver of any
provision or right. Neither the course of
conduct between parties nor trade practice will
act to modify any provision of this Agreement.
In other words, Susie runs SuSe and sshd on her
home box so she can use, e.g., scp from work to home and back. Now, that *could* be seen as an illegal server (although there could be a loophole due to the fact that the TOS only bans servers used by "others"). But in fact, they would never
bother with this. On the other hand, Ned is a
napster user who is not only sucking up all the
local bandwidth but arguably sharing MP3s in a way that could be construed to involve copyright infringement. They will probably tell him to knock it off.
The most worrisome point of this TOS is one that I haven't seen mentioned yet:
8.1 You understand and agree that Sprint's
network gathers information about internet
usage such as the sites visited, session lengths,
bit rates and number of messages and bytes
passed. Sprint uses this information in the
aggregate. Sprint may share this aggregated
information with other parties from time to
time. Sprint will not use or disclose any
personal identifiable information regarding
internet usage unless compelled by a court
order or subpoena or [sic]You consent to the use or disclosure [sic]or to protect Sprint's Services and facilities.
I've marked possible typos with [sic] here. One problem here is that there is arguably
never any good (non-marketing) reason to collect
"personal identifiable information" in the first
place, since this is the kind of thing that is ripe for subpoena. Indeed, most corporations are
learning the lesson that they should do things like forcibly delete email and log information pretty rapidly, just so that third parties won't have anything to go after. Now, it's possible that Sprint never really does store personal identifiable information except email, but they don't say that.
Problem two is that even if they only collect information about web-browsing behavior in the aggregate, they (or other marketers) can often
have other information about you that can be used
to do an alarmingly good job of determining who
surfed where...and targeting you for directed marketing campaigns in the process. Now, there may be very little escape from the Database Nation these days, and you can argue that at least they
are being frank, but this kind of thing still gives me the creeps.
Jesus H Christ, am I the only one that doesn't give two monkey turds about seeing big, huge massive, eye-candied, screen realestate hogging icons representing all of my files? I mean come on, some of those screen shots show all of 4 files in a window that is presumedly running in at least 800x600 resolution. This is user friendly? I just want to find and launch my files.
Well, a few things come to mind. The first one
is that the Eazel folks are trying to design in the idea of expertise in the interface. So most of the screenshots I've seen, some of which are really silly, are
from the novice view. I haven't seen much of the
so-called Intermediate and Advanced views, but I suspect that they use available screen real estate
much differently.
It's also clear that Eazel is way easier to customize than current desktoppy environments; I've never seen a screen shot that didn't let you
view as things other than icons.
They're also claiming, and this will be the interesting part, that they can also give you easy access to content and attribute-based views of your file system. OK, so I've
got 166 pdf files, 496 postscript files, and 75 other EPS files in my personal account alone. The problem is, of course, that they're scattered all over the directory hierarchy, and it can be tricky to find (or remember) exactly where any one document is. It would be
great to have a view of my files that would bring together all the "postscripty" files, or just the
non-graphical postscripty files, or all postscripty files that belong with research projects, or what have you. Now, you can do that with a standard unix file system through clever, deliberate, and pre-planned thinking about your directory structure (and maybe the use of symbolic links), but almost nobody does this. And, if they do, they then find the nth plus one way they *really* want to look at stuff, but which is difficult with their pre-calculated layout and the available tools.
And there really are situations where a standard hierarchical view of my files just gets in the way. Now that most of us
have thousands of little twisty files hanging around, there has to be an answer somewhere. One
idea is to change our filesystems to something more flexible (based perhaps on a relational model). Another idea, which I take to be similar to Eazel's, is to layer (and in some cases, infer)
an additional attribute database over the file system so that you can see what you've got in the way most relevant to you at the moment. Now, it's possible that they forgot to make this part sufficiently scriptable for the common geek, but I severely doubt it.
So I'm not sure whether Eazel's stuff will be the bee's knees, but I'm encouraged that they've figured out that different views for different purposes and users is the key concept to implement, rather than (merely) nicer looking icons or pin-striped menubars.
I've always felt that the solution to all the language wars would be if people learned different programming languages preferrably from different programming paradigms. A Perl bigot wouldn't go on and on about how truly OO Perl was if he knew about SmallTalk and neither would a C bigot rhapsodize about C's text-handling abilities if given Perl.
I'm all for people learning different kinds of languages, but not (just) to prevent them from saying silly things. Silly people will say silly things no matter how well-informed they are. (OK, so the subject of their silliness may move around a bit.) And it's certainly not to educate them about why language X, which they used productively for some time, is inferior to language Y in some regard (although that sounds a bit more like your argument). People should try different things
because new experiences teach them about new problems and new ways to solve problems old and new. Even if they continue to use their favorite tools rather than the ones that a first principles analysis might suggest.
So, along the lines of this particular thread, it's a bit amusing to note that one of the best
books on object-oriented programming I have ever
seen is Damian Conway's Object-oriented Perl book. And the reason for this is not because it's a perl book or even a book on OO, but because it's a book that shows you a mind-bogglingly large number of ways to look at problems in ways that can be usefully construed as being "object-oriented". And, from that, you get a much better idea about what objects really are. And it's fun. It turns out that the author, a CS professor who does research and teaching on OO things, found perl to be useful for this book due to its chameleonality.
Yes, a person who knows more about Smalltalk and C++ and Self and Python and the rest of it than do most of us found that a good language to talk about all of these things was Perl, although it may well have been not the "best" language to use for any of them.
When it comes to type systems, I like Scheme; it has a nice, clean type system. Perl can accomplish all the same nifty tricks as Scheme can, (it has closures, yay!) but you have to at least use a lot of references, and the syntax can get pretty grotty.
OK, this is not a flame. But I think your post
basically proves why Mark Jason Dominus really
has a point in his article, and then some.
You start with a compliment about Scheme and it's type system, then the very next statement is in the spirit of why something else sucks. Now, granted, you were waaaay gentler than many people are, but the kid gloves get dropped by the
third iteration on most of these threads.
Your post also points out a second real problem with most advocacy debates: both sides define the terms however they please, often indiosyncratically. And then they proceed to talk past each other until somebody points out that this is what is happening, then they argue about whose definitions of terms are
more appropriate...you get the picture. In this
particular case, I would point out that somebody who was interested in type systems might be pretty confused by your description of scheme and its type system, since scheme is often considered to be basically untyped (certainly compared to ML, which comes up in the target article).
I think the only useful kind of advocacy probably takes the form of seeing what the language can actually do. I have learned things when,
for example, I saw that two languages differ in their solutions to one kind of problem, and that one or the other might make better sense in the context of other problems you had to solve. Dominus goes on to point out that perl, notably but not uniquely among currently developed languages, owes many or most of its features to some kind of agglutinizing proccess like this.
Re:No more deja.com/usenet
on
Deja.com Vu!
·
· Score: 2
Unfortunately, I don't see how it can be made profitable in these post-banner-ad days. But a public service like this needs to be maintained! All I can pray for is that an internet philanthropist like Brewster Kahle decides to
buy up the archive and put it online at a loss. If I had the money I'd do it myself. (And no, I wouldn't charge for it - I consider that highly inappropriate since the postings were made freely)
The problem with any idea involving internet philanthropy is that it can be horrifically expensive if you give it enough bandwidth, but reasonably useless if you don't.
My personal guess is that something like dejanews is most valuable to a successful company in the
searching and indexing world. Actually, the specific candidate I have in mind is google. I think it's pretty obvious that the google model for rating/ranking the usefulness of content could be applied to Usenet archives in a big way (the biggest challenge being to recognize and deal appopriately with flame wars). It's also the case
that if people "knew" or "believed" that dejanews URLs would be permanent in some sense, that they'd use and link to them as they would to any other net resource, and the usefulness of the whole enterprise would go way up. So you type something like "linux kernel usb support" into google, and then you'd get back not just a few random-ish web pages, but the key usenet threads on the subject.
Heck, now that yahoo is growing more dependent on google to be it's web-indexing device, it should become really clear to google that they really, truly want to do this.
At least, I can hope...
Re:A good step in the right direction
on
Deja.com Vu!
·
· Score: 2
revenue of $922,000 in the first quarter
That kind of revenue stream could pay for colo space, bandwidth, and the salaries of a handful of techies and programmers.
You have the right idea, but very possibly the wrong numbers (these data were from 1998). Now
that we have better data on user behavior, and a not-so-booming economic situation to play with,
It wouldn't surprise me if deja.com (or even dejanews.com:-)) could only depend on half or a third of that income stream. Then I think the
cost situation gets a lot more tricky.
With any luck, they will realise there is a market for searching usenet archives going back 15 years, and set up a parallel pay-for-performance system for corporate groups. If you were the head of a large programming group, would you pay for a local copy of deja's DB, covering comp.* and alt.* from 1985 to a few months ago?
There is *some* market for this, sure, but less than you might expect. I think it was Henry Spencer who pointed out that, if you extrapolate from the data we had, the truly most useful and popular post on Usenet is the post you are about to make (or am I confused?). Now, I *loved* the fact that I used to be able to go back in time using the dejanews archive and find
out how much of a weenie I used to be in the early 90s, but I'm afraid I can't (yet) impute a beefy profit stream to Usenet nostalgia.
They could also sell their text archiving and indexing system to large companies, the same as google, ask jeeves, and others are doing.
So why don't they do this? To paraphrase the Old Prospector in Toy Story 2: Two words: Remarq Dotcom. It's very possible there's still some money in this kind of
thing, but I'm not sure.
would our eyes accept something so radically different than what we're used to (horizontal scanning)?
Why not? First of all, does your eye accept looking at a monitor sideways?
I think you have a misconception of how CRTs work... Yes they scan sideways but they scan up and down too...side to side, then advance a line.
A yup. But just to amplify this a bit, the CRT miracle (a steady visual percept over the entire screen caused by a weird serial scan across the phosphors of the screen) is caused by two things.
The first (and easiest to understand) is the fact that phosphors don't decay instantaneously, so that wouldn't (in principle) need to scan the whole screen at once on a CRT. The second (and more important) thing is that the human retina is
a s-l-o-w beast, with the effect that our "instantaneous" percept is actually the average of
activity in a surprisingly broad temporal window.
Now, in the bad old days, you could easily get "flicker" in a monitor display, especially a cheap
color one, because the time required to draw the
screen was just long enough for phosphors to decay
noticably *and* just long enough so that the temporal averaging done in the retina didn't help you. But even that was only true in your rod-dominated peripheral vision (where the aptly-named "flicker fusion rate" is higher). And
the worst cases were largish screen sizes with huge dot pitches and interlaced displays.
But those days are essentially gone now that we
can easily repaint a screen at 70 Hz or even 120 Hz. Now, things are a bit trickier for large back projection displays, since the screen has no phosphor, and you're going strictly on retinal "hang time", but this technology is now well under
control for modest resolutions. What this new technology promises is an easier way to scale up resolutions to match full HDTV richness, which is, of course, a Good Thing.
We don't need more precision in desktop computing. 32 bits is arguably plenty for any floating point number in use on a desktop.
I respectfully disagree. I'm virtually certain that if you administered truth serum to application writers who know anything about numerics, they would swear up and down that they never want to deal with anything other than IEEE 754 standard 64-bit double precision numbers, and are only forced to do so due to dorky efficiency
concerns with stock (commodity) hardware.
The legacy of backward compatibility (which amounts to backward capability in many situations) is one of the biggest barriers to advances in consumer and desktop machines at this time. An interesting (and possibly vital) point about Free and/or Open software is that it's far
quicker and easier to adapt older applications to new platforms because enough of the affected users are empowered to improve and change the legacy apps.
One other nit about about the need for more precision and floating point: for slightly more than historical reasons, there is still at best
squeamishness about using FP arithmetic for certain financial calculations, and a 32-bit unsigned integer quantity is only able to represent values in the range of the milli-Gates or milcro-GDP...
[an analyst writes that] Red Hat "has $320 million in cash on hand, that it consistently meets quarterly revenue expectations, and that its gross margins are improving."
All of which are true. But until RHAT starts making money, it's still not necessarily a good investment.
Well, I both agree and disagree with your statement. In actuality, you should have said "when RHAT resumes making money".
People tend to forget this, but RedHat *did* earn
$.04 per share one quarter last year, and *did*
earn money from time to time before their IPO.
If you look at all the quarterly data, it does appear that they're far closer to having actual earnings in 2001 than many other (fallen) IPO littermates. Now, having said that, I think the
last quarterly earnings report, due out any day now, is *really* important. If they beat the analyst estimates, possibly by breaking even last
quarter, then people will feel much better about
their prospects again.
It's a better buy at $6.50 than at $150. But that doesn't make it a good buy.
With $42M in revenue, you're still paying about 25 years' worth of revenue for every dollar's worth of stock you buy.
The comparables for the rest of the sector are 12 years, and for the rest of the market, two years.
The other day I tried to figure out what price I'd buy Red Hat at, and the answer was about $4 per share until I knew what the last quarter was really like. Given the lack of earnings warnings,
I suspect they finished somewhere between -.02 per
share to break even. But now I *suspect* there will be some charge against earnings in the next quarter for the consolidation of offices (3 offices and 20 people got trimmed). However, this should be put into perspective: I think Red Hat should be making about $5 million in interest on their cash this last quarter, and next year they could make about $20 million. That alone is enough to paper over a lot of sins, so if their sales revenue grows at anything like the rate I would expect, they will be well into the black by
this time next year.
Annoyingly, I suspect their stock will remain just a tad over-priced for the foreseeable future.:-(
There's a really interesting overview article on color vision in the Feynman Lectures, volume I. It includes typical spectra for R, G, and B dyes. If I recall right, R and G are actually rather similar spectrally, with somewhat broad humps in the long end of the spectrum, while the B dye has a very different spectrum with a sharp peak near the short end of blue.
Most of this post was essentially correct, but I just wanted to amplify this part of the message.
Yes, if you look at the spectral sensitivities of
red, green and blue cones (or, strictly, their dyes), blue is many nanometers shorter in wavelength than the difference between red and green. But to test your understanding of how
color "works" at the retinal level, the key question should be: where does "yellow" come from?
The answer, of course, is from the additive contributions of both red and green cones; indeed, when you look at the sensitivity curves, you can see that the response to "yellow" should be larger than either green or red. And,
it is. Visual acuity is actually slighly better
for yellow than for any of the primaries (think shooter's glasses). Now, having said that, I should point out that blue is a special weird case, since the blue cones have a much more limited distribution on the retina than do red and/or green cones.
And, having said all of this, the most amazing thing about color vision (in my opinion) is not
what happens at the retina, but what happens in
the cortex, apparently in area V4. That's where
the very hard problem of color constancy (aka "discounting the illuminant") is solved in a manner studied at great length by Edwin Land, who really would be every geek's hero if only he were better known.
as the amusing recollection of the "vibrating" footbal field.
Was anyone who had one of these ever able to get the QB to pass or the kicker to kick?
I know I couldn't...
Well, *I* could get the QB to pass, as could my
little brother.
The only difference was that his QB
always threw incomplete passes, while mine always
threw interceptions. Indeed, most of the points scored in our games happened when my QB threw an
interception to his "magic guy", if you know what
I mean.
On the bright side, the rest of the points in our games happened when his lame ass kicker dribbled
a kick on the ground to my "magic guy", or the
occasional safety happened. I think I did kick
a field goal, once.
Way better than this was the rods-and-slots hockey
game, where the "players" were these cardboard things that snapped onto the "stick" bases; you could get different teams in different "uniforms"... On the down side, no magnets were
involved.:-)
Secondly, the margin for Gore's victory in the so-called popular vote is something like 0.3%, well within of the margin of error.
Others have said this, too, but I'm afraid that
I don't know what they're talking about. If
you would like to see some careful work on margins of victory (especially in Florida, surf on over
to Professor Rusin's paper on this matter. After reading this, you should be clear on the concept
that a margin of 200,000 votes out of 100,000,000
cast is highly unlikely to be reversed on any kind of more careful recount unless the original count suffered from some systematic bias affecting one candidate or another. You will then also find out how likely it is that, e.g., the Florida election outcome would be reversed by recounting ballots in counties where you know there to be systematic effects. Basically, Florida would still be close to a statistical dead heat in the situation where we have the totals we have now and the knowledge that there are thousands of untabulated ballots in counties known to prefer one candidate or another.
The questions here are political rather than mathematical, which is why the situation is so awkward for both sides.
So not only can we say that recounts (like those we've had in florida) could result in shifting of numbers, but we can also reasonably presume that the slightest change in behavior of either of the candidates could have overcome that margin (i.e., under a popular vote).
This sort of discussion went down at the University I work at a couple of years ago.
That time period is now called by everyone (even the upper level of University management) the Email Wars.
So, is this about the University of Missouri Email Wars, or is this Yet Another University Email War (YAUEW!)? I ask, because your whole message sounds eerily familiar.:-)
At Mizzou, we have thousands and thousands of users who have either "missouri.edu" addresses (faculty and staff) or "mizzou.edu" addresses (lowly students). Our system is unstable, slow,
and almost certainly not worth it despite an
IT squad that is generally smart, responsive, and
eager to please. Frankly, I would never have
thought I could care less what the email server
was that I was using, but now, alas, I know better.:-(
Slashdot Mirror
The most reasonable way to read this comment is that Big Science costs Big Money. And that is basically true. And there is no doubt that certain truly Huge projects, like the HGP, cost a lot of money. What is quite interesting about the HGP is the fact that there was surprisingly little disagreement that the project was worth doing, and that a public consortium should do it. And this was yet another big reason why Celera annoyed so many people.
But even though the HGP was as Big a science as it gets, the thing that really takes your breath away is how Small some of the most crucial ideas were; ideas that ending up speeding the whole enterprise by years. This, for me, is one of things that makes Big Science worthwhile: there is so much Small science waiting to get out. Getting any grant to do anything approved these days is an exercise in masochism. It's much easier to burrow into a nice, warm, big project than do the small stuff by yourself. I think this is why much of the really big stuff ends up getting done either by accident, or by somebody willfully doing something other than what they were supposedly getting paid for.
Of course, Big Science (whether public or coprorate) doesn't always end up appreciating the nuggets of innovation that pan out from time to time. Shotgunning the genome was something that the HGP didn't really think was going to fly, or at least fly so fast, so when push came to shove, Ventner walked out the door.
In retrospect, probably a pretty good move.
I'll actually hazard a guess here that the answer is "no". If we're talking about Deja, it looks to me like that plane was going down so fast (with both engines burning, noxious gasses filling the cockpit, etc.) that the cost-effectiveness of keeping up their web presence (which was not *that* cheap) was about zero. They knew they were going to assimilate the database, that the re-structuring required was massive, that their own current archive was possibly in better shape than the one belonging to a not-really-going-anymore concern, and that there was no point in waiting to kill off something they had no interest in supporting in the future.
I'll grant you that this sucks a bit, but I'll be willing to bet at least a little bit that the rapid crash-and-burn had at least as much to do with deja's deteriorating situation as it did with Google's arrogance/lack-of-planning/whatever. The Google press release on the Deja acquisition basically states that they acquired all of the "significant assets" of deja.com, which is a fancy way to say that this was not a merger, but at best a firesale, and that the insignificant assets are probably stuff like your former deja account. :-(
This is a question for...PSYCHOLINGUIST MAN!
To be completely serious, this is a perfectly grammatical sentence. Indeed, I think it would make my Good Buddy Robert Kluender beam with joy. Now, is this kind of thing a piece of cake to parse? No way: it has what we experts call an unbounded wh-dependency. Indeed, our willingness to torture^H^H^H^H^H^H^Htest undergraduates with stuff like this is why we make the big bucks.
Now, to prove to you that this sentence is legit, consider the following:
Does this help any? Now, the real interesting question is why people would tend to say (4) above as What do you think the odds are of Google acquiring such data? But I have office hours in five minutes, so that question will have to wait for another day. ;-)
To make this just remotely related to the topic of search engines and Usenet, I'll point out that long distance dependencies like this one are the kind of thing that can make it infuriatingly difficult to use easy cues like "lack of proximity" to decide that two search terms are truly unrelated to each other. Unfortunately, solving this one requires you to parse natural language as it is used on Usenet, which is truly a frightening thought.
In the grocery store. The people who were the big push in bringing you Universal Product Codes, instant coupons based on what you bought, and, let's face it, the entire shopping cart metaphor on which the internet is based (:-)), have also brought you, in many markets, the Value Card. (OK, sometimes it's a key-chain or dongle these days.) The only thing you need to do to get one is to fill out the handy form...and then, from then on, everything you buy from there is recorded as having been bought by you. Everything from avacados to zesta saltines, and from birth control pills to Prozac.
It's a marketer's dream.
Now, you do, of course, get something for your trouble: sales and price breaks that, in my experience, never amount to anything more than you could have gotten from the same grocery store before they had the "Value" program, or more than you could get from a competing grocery store that did not have a "Rewards" program (or whatever they might call it). So let's calculate the value of your privacy, as estimated by the market:
Hmm, that looks like about Zero, to me. Now maybe it's possible that when all stores have these programs, and many people are essentially forced to join one to get the kind of prices they used to get without selling all of their privacy, it's possible then that we would see market competition between grocery stores concerning what you get for your privacy rights. Or not. What we might more likely see is a vast expansion of the "instant coupons" idea, where your price and my price for the same good can vary depending on the rest of our buying habits, and which brands are willing to shell out a little bit more in price breaks for their least loyal customers.
We lived in California for just over 7 years, and during the period when they finally got the utility de-regulation plan enacted. And, I swear, I could never figure out why in the world most people wanted to do this. The utility situation in California (and you could include water in this as well) is a truly classic example of where markets will have a rougher time, because virtually every major change or transaction involves an external party who would not be taking part in the transaction if it were a conventional market.
The building and siting of power plants, for example, is a problem that comes up almost everywhere, but the costs to other parties of putting them wherever the grid thinks them most efficient are vastly greater in California than in most places. California generates a huge amount of it gross state product from the fact that it *is* California, the idyllic (-looking) paradise. In California, a huge determinant of the value of any piece of land is, to be quite frank, the view, and the clarity of the air and water, which are both kinds of "rights" that are extremely hard to deal with if the holders aren't a party to the transaction.
Now, what actually happened was the way people became party to these kinds of transactions was through the political system. Most specifically, the PUC and their right to regulate the placement and operation of power plants and the prices that could be charged. In return for this power, utilities were essentially granted a guaranteed rate of return (like most other places), which is a boring but perfectly profitable way to do business. Yes, there is no doubt that there were inefficiencies in that system, but, because it was political, all the affected parties were involved, and everybody could play. This was the reason why I find it hard to consider the PUC to be truly a central planning agency: everybody could and did put tremendous amounts of pressure on the PUC to have things their way, but always got compromise solutions. The problem, of course, was that some users believed that they could get better prices through a different system, and that it was worth the huge monetary cost of pushing hard on the political system in the usual fashion. This would have been okay, except that the compromise solution that was reached had some spectacular bugs in it that weren't fully appreciated at the time, but which could be (and have been) exploited to the hilt by power suppliers, who were (by law) completely separated from the utility companies themselves.
So, when I see what happened, I don't think of it as a market failure or a central planning failure, but as a political failure of the type that so very often crops up in California. Why California more than most other places? I think it is because the stakes are so much higher there than other places, and because there are incredibly strong regional political differences that make the place basically ungovernable. I cannot possibly imagine something like this happening in, say, Iowa. Or in a California that consisted of two or even three independent states.
I respectfully disagree completely. :-)
Seriously, truly persistent document storage with automatic infinite undo, the ability to play with any version you want, automatic indexing and a built-in relational view of all your other documents...I'd be in heaven. The true value of persistent state isn't just saving a few keystrokes here and there. The big win isn't in the data; it's in the metadata, the linkage, the automatic indexing and all the stuff that is truly tedious to do yourself.
In the olden days, this didn't seem practical, but next year I can probably get a 200 gig hard disk for under $300. Which probably only means more empty disk space unless somebody figures out that it's okay to blow a meg of storage on every text file I ever create if that's what it takes to provide all the real services people want in a file system (or document system). The capacity of that 200 gig disk is convincingly larger than the entire unindexed capacity (in bits) of the human memory that will serve you so conveniently for decades, and may (finally) be big enough to hold the index, too.
Now, lest any of you believe I'm condoning some kind of device for idiot, keep in mind that I've said nothing about the beautiful query language I have in mind. The use of that is what will separate the experts from the novices...
Like, wow; something finally sunk in for me. As far as what aspect of Microsoft is a wonder of the economic world, I would have though the answer was: that absurd P/E ratio. It's now down to 33 (or up to 33 if you only consider the last few months). If Microsoft, with complete market domination, can only grow as fast as the PC industry, then their P/E ratio really shouldn't be greater than the PC industry (check), and the PC industry, now that we're nearing saturation, can't grow much faster than the economy at large, which is worth a P/E of like 22 these days. So where is the extra growth going to come from? If their margins are really 24%, it's not likely to be from an increase at the margin, and it's not coming from an increase in market share, and it's not coming from an increase in market size. Looks pretty bad to me. So if we were to reduce their margin to like 6% (nothing to sneeze at), hold their P/E to be equal to the market average, and put the market P/E back at 15 where it has tended to be, I get a target price for MSFT at...
About $7.50 per share.
I think that's a bit low, but it get me rather worried about what's going to hit the fan when and if the market really comes back down to earth.
This has been mentioned as a legal possibility, but it has the following important limitation: Any settlement has to be approved by the trial court judge. It wouldn't be inconceivable that Judge Jackson would approve a settlement that would end the Appeals process (remember: he's a conservative Republican and a Reagan appointee), but given the contempt for his court that he detected in Microsoft's conduct of the original case, I wouldn't bet on him giving up too much without a pretty thorough confession of wrong-doing by the "we did nothing illegal" MS establishment.
You give me the giggles. :-)
Seriously, you sound very confused to me. Your hand-baked approach to putting physics on the web didn't work so well, because, as you point out, there was no way to separate form from content that way. But then you slam on TeX for, well, trying to enforce that useful separation.
The second objection is (I think) to the way that TeX actually dares to compile your document, report errors, and work essentially in batch mode. Yes, that can be infuriating at times, but the advantage is that, at the end of the day, your work really could be device independent. Giving up that is giving up the farm, in my opinion.
The first objection is the one that I found most amusing, though. Like, so what if only one person in ten thousand ever learns to use LaTeX? Many fewer people than that ever create anything that needs it. Heck, only a few people in a hundred ever create anything much at all, and it's a given that fewer people know how to use tools than know how to appreciate the results of tool use.
We are the geeks; the people who use tools.
The people who use tools to make tools.
The people who use tools to design tools to make tools. We are the lords of the Shell, the emperors of Perl, we dominate the DOM.
You're totally right that bitmaps aren't the way to do equations (or lots of other things). I can agree that LaTeX isn't perfect or even that close. But it was the tool that showed that not everything had to be set by hand, that math could be free, that something like MathML would eventually become a worthy successor. Worthy, that is, if we're clever enough to design a useful input device for the stuff...
This part is easy. With X, you can run apps remotely and display them where you are. So, for me, anything I want to run on a Linux box at work but view at home is such an app. Now, I expect that's not the answer you were expecting, but if X doesn't immediately suggest to you the possibility of remote apps, you're missing the point.
Nope; that part was easy.
Well, I'm married with two kids. My wife vastly prefers the Mac interface, and an iMac was decent enough looking and quiet enough to put out in a public place, saving additional bucks for one of those "hide the ugly PC" computer thingies. Now, I do kick myself for not waiting another 6 months and getting the iMac I got for $500 less, but that's the PC life...
No, that was still pretty easy.
Now the funny part of this is that even though I live in Columbia, Missouri, a metro area with a population of 130,000 or so, I already know a handful of people like this. But so what? Why should I care how many other people use computers like I do? In the bad old days, that might matter because software was expensive and and closed source; not these days. I can have a Mac, a BSD box, a wireless network set-up, a DVD player, and all the comforts of a quiet-to-silent PC for a price I was more than willing to pay.
Nobody can tell me I'm not happy.
Me too. But, not being nearly wealthy enough to afford so-called "point of service (POS)" insurance, we make do with a so-called "Health Maintenance Organization (HMO)" which has all the charms of the Canadian system without some of the advantages.
Maybe it's just me, but isn't it a good thing to get the pus-o-matic heroin addicts out of the public waiting area as quickly as possible? I mean, even an HMO will do that. Take my five year old son; no, he's not a disgusting heroin addict, but he did today have a glorious case of chicken pox, so the "pus-filled" part applies. Anyway, the HMO had him in and out of that room in under 5 minutes. But next time I have to go in there, I think I might try the old Canadian "heroin needle in the arm trick" to get faster service...
Well, down here it seems like the market demands a 40% Backstreet Boys Content (BBC); sometimes it's unpleasant to get what you ask for.I would feel your pain, but, hey, you also get to hear Crash Test Dummies! And, uh, Rush? And Gordon Lightfoot and...uh, you mean to tell me that the music up there looks just like the godawful Air Canada in-flight music program? If so, I think you could challenge this as a human rights abuse, right? :-)
OK, now that you mention it, what *is* up with that? Are courses in lighting design just not offered north of the border (or in many European countries)? Does this have something insidious to do with the metric system? (Nope; Japanese TV lighting is superb.)
Then don't come to Missouri; here you have the Canadian winters plus the special treat of those dew-points-in-the-high-70s days that we call "summer". We cheat cheat death 8 months out of the year by daring to leave our climate-controlled subdivision houses and SUVs.
Yup. This particular TOS is pretty much designed to make sure that they can enforce something at some time rather than everything at every time. Specifically, they mention that:
In other words, Susie runs SuSe and sshd on her home box so she can use, e.g., scp from work to home and back. Now, that *could* be seen as an illegal server (although there could be a loophole due to the fact that the TOS only bans servers used by "others"). But in fact, they would never bother with this. On the other hand, Ned is a napster user who is not only sucking up all the local bandwidth but arguably sharing MP3s in a way that could be construed to involve copyright infringement. They will probably tell him to knock it off.
The most worrisome point of this TOS is one that I haven't seen mentioned yet:
I've marked possible typos with [sic] here. One problem here is that there is arguably never any good (non-marketing) reason to collect "personal identifiable information" in the first place, since this is the kind of thing that is ripe for subpoena. Indeed, most corporations are learning the lesson that they should do things like forcibly delete email and log information pretty rapidly, just so that third parties won't have anything to go after. Now, it's possible that Sprint never really does store personal identifiable information except email, but they don't say that.
Problem two is that even if they only collect information about web-browsing behavior in the aggregate, they (or other marketers) can often have other information about you that can be used to do an alarmingly good job of determining who surfed where...and targeting you for directed marketing campaigns in the process. Now, there may be very little escape from the Database Nation these days, and you can argue that at least they are being frank, but this kind of thing still gives me the creeps.
Well, a few things come to mind. The first one is that the Eazel folks are trying to design in the idea of expertise in the interface. So most of the screenshots I've seen, some of which are really silly, are from the novice view. I haven't seen much of the so-called Intermediate and Advanced views, but I suspect that they use available screen real estate much differently.
It's also clear that Eazel is way easier to customize than current desktoppy environments; I've never seen a screen shot that didn't let you view as things other than icons.
They're also claiming, and this will be the interesting part, that they can also give you easy access to content and attribute-based views of your file system. OK, so I've got 166 pdf files, 496 postscript files, and 75 other EPS files in my personal account alone. The problem is, of course, that they're scattered all over the directory hierarchy, and it can be tricky to find (or remember) exactly where any one document is. It would be great to have a view of my files that would bring together all the "postscripty" files, or just the non-graphical postscripty files, or all postscripty files that belong with research projects, or what have you. Now, you can do that with a standard unix file system through clever, deliberate, and pre-planned thinking about your directory structure (and maybe the use of symbolic links), but almost nobody does this. And, if they do, they then find the nth plus one way they *really* want to look at stuff, but which is difficult with their pre-calculated layout and the available tools.
And there really are situations where a standard hierarchical view of my files just gets in the way. Now that most of us have thousands of little twisty files hanging around, there has to be an answer somewhere. One idea is to change our filesystems to something more flexible (based perhaps on a relational model). Another idea, which I take to be similar to Eazel's, is to layer (and in some cases, infer) an additional attribute database over the file system so that you can see what you've got in the way most relevant to you at the moment. Now, it's possible that they forgot to make this part sufficiently scriptable for the common geek, but I severely doubt it.
So I'm not sure whether Eazel's stuff will be the bee's knees, but I'm encouraged that they've figured out that different views for different purposes and users is the key concept to implement, rather than (merely) nicer looking icons or pin-striped menubars.
Ah, but just imagine the reply, if it had been sent in 2001...
Well, it *might* have happend that way...
I'm all for people learning different kinds of languages, but not (just) to prevent them from saying silly things. Silly people will say silly things no matter how well-informed they are. (OK, so the subject of their silliness may move around a bit.) And it's certainly not to educate them about why language X, which they used productively for some time, is inferior to language Y in some regard (although that sounds a bit more like your argument). People should try different things because new experiences teach them about new problems and new ways to solve problems old and new. Even if they continue to use their favorite tools rather than the ones that a first principles analysis might suggest.
So, along the lines of this particular thread, it's a bit amusing to note that one of the best books on object-oriented programming I have ever seen is Damian Conway's Object-oriented Perl book. And the reason for this is not because it's a perl book or even a book on OO, but because it's a book that shows you a mind-bogglingly large number of ways to look at problems in ways that can be usefully construed as being "object-oriented". And, from that, you get a much better idea about what objects really are. And it's fun. It turns out that the author, a CS professor who does research and teaching on OO things, found perl to be useful for this book due to its chameleonality. Yes, a person who knows more about Smalltalk and C++ and Self and Python and the rest of it than do most of us found that a good language to talk about all of these things was Perl, although it may well have been not the "best" language to use for any of them.
OK, this is not a flame. But I think your post basically proves why Mark Jason Dominus really has a point in his article, and then some.
You start with a compliment about Scheme and it's type system, then the very next statement is in the spirit of why something else sucks. Now, granted, you were waaaay gentler than many people are, but the kid gloves get dropped by the third iteration on most of these threads.
Your post also points out a second real problem with most advocacy debates: both sides define the terms however they please, often indiosyncratically. And then they proceed to talk past each other until somebody points out that this is what is happening, then they argue about whose definitions of terms are more appropriate...you get the picture. In this particular case, I would point out that somebody who was interested in type systems might be pretty confused by your description of scheme and its type system, since scheme is often considered to be basically untyped (certainly compared to ML, which comes up in the target article).
I think the only useful kind of advocacy probably takes the form of seeing what the language can actually do. I have learned things when, for example, I saw that two languages differ in their solutions to one kind of problem, and that one or the other might make better sense in the context of other problems you had to solve. Dominus goes on to point out that perl, notably but not uniquely among currently developed languages, owes many or most of its features to some kind of agglutinizing proccess like this.
The problem with any idea involving internet philanthropy is that it can be horrifically expensive if you give it enough bandwidth, but reasonably useless if you don't.
My personal guess is that something like dejanews is most valuable to a successful company in the searching and indexing world. Actually, the specific candidate I have in mind is google. I think it's pretty obvious that the google model for rating/ranking the usefulness of content could be applied to Usenet archives in a big way (the biggest challenge being to recognize and deal appopriately with flame wars). It's also the case that if people "knew" or "believed" that dejanews URLs would be permanent in some sense, that they'd use and link to them as they would to any other net resource, and the usefulness of the whole enterprise would go way up. So you type something like "linux kernel usb support" into google, and then you'd get back not just a few random-ish web pages, but the key usenet threads on the subject.
Heck, now that yahoo is growing more dependent on google to be it's web-indexing device, it should become really clear to google that they really, truly want to do this.
At least, I can hope...
You have the right idea, but very possibly the wrong numbers (these data were from 1998). Now that we have better data on user behavior, and a not-so-booming economic situation to play with, It wouldn't surprise me if deja.com (or even dejanews.com :-)) could only depend on half or a third of that income stream. Then I think the
cost situation gets a lot more tricky.
There is *some* market for this, sure, but less than you might expect. I think it was Henry Spencer who pointed out that, if you extrapolate from the data we had, the truly most useful and popular post on Usenet is the post you are about to make (or am I confused?). Now, I *loved* the fact that I used to be able to go back in time using the dejanews archive and find out how much of a weenie I used to be in the early 90s, but I'm afraid I can't (yet) impute a beefy profit stream to Usenet nostalgia.
So why don't they do this? To paraphrase the Old Prospector in Toy Story 2: Two words: Remarq Dotcom. It's very possible there's still some money in this kind of thing, but I'm not sure.
A yup. But just to amplify this a bit, the CRT miracle (a steady visual percept over the entire screen caused by a weird serial scan across the phosphors of the screen) is caused by two things.
The first (and easiest to understand) is the fact that phosphors don't decay instantaneously, so that wouldn't (in principle) need to scan the whole screen at once on a CRT. The second (and more important) thing is that the human retina is a s-l-o-w beast, with the effect that our "instantaneous" percept is actually the average of activity in a surprisingly broad temporal window.
Now, in the bad old days, you could easily get "flicker" in a monitor display, especially a cheap color one, because the time required to draw the screen was just long enough for phosphors to decay noticably *and* just long enough so that the temporal averaging done in the retina didn't help you. But even that was only true in your rod-dominated peripheral vision (where the aptly-named "flicker fusion rate" is higher). And the worst cases were largish screen sizes with huge dot pitches and interlaced displays.
But those days are essentially gone now that we can easily repaint a screen at 70 Hz or even 120 Hz. Now, things are a bit trickier for large back projection displays, since the screen has no phosphor, and you're going strictly on retinal "hang time", but this technology is now well under control for modest resolutions. What this new technology promises is an easier way to scale up resolutions to match full HDTV richness, which is, of course, a Good Thing.
I respectfully disagree. I'm virtually certain that if you administered truth serum to application writers who know anything about numerics, they would swear up and down that they never want to deal with anything other than IEEE 754 standard 64-bit double precision numbers, and are only forced to do so due to dorky efficiency concerns with stock (commodity) hardware.
The legacy of backward compatibility (which amounts to backward capability in many situations) is one of the biggest barriers to advances in consumer and desktop machines at this time. An interesting (and possibly vital) point about Free and/or Open software is that it's far quicker and easier to adapt older applications to new platforms because enough of the affected users are empowered to improve and change the legacy apps.
One other nit about about the need for more precision and floating point: for slightly more than historical reasons, there is still at best squeamishness about using FP arithmetic for certain financial calculations, and a 32-bit unsigned integer quantity is only able to represent values in the range of the milli-Gates or milcro-GDP...
Well, I both agree and disagree with your statement. In actuality, you should have said "when RHAT resumes making money". People tend to forget this, but RedHat *did* earn $.04 per share one quarter last year, and *did* earn money from time to time before their IPO. If you look at all the quarterly data, it does appear that they're far closer to having actual earnings in 2001 than many other (fallen) IPO littermates. Now, having said that, I think the last quarterly earnings report, due out any day now, is *really* important. If they beat the analyst estimates, possibly by breaking even last quarter, then people will feel much better about their prospects again.
The other day I tried to figure out what price I'd buy Red Hat at, and the answer was about $4 per share until I knew what the last quarter was really like. Given the lack of earnings warnings, I suspect they finished somewhere between -.02 per share to break even. But now I *suspect* there will be some charge against earnings in the next quarter for the consolidation of offices (3 offices and 20 people got trimmed). However, this should be put into perspective: I think Red Hat should be making about $5 million in interest on their cash this last quarter, and next year they could make about $20 million. That alone is enough to paper over a lot of sins, so if their sales revenue grows at anything like the rate I would expect, they will be well into the black by this time next year.
Annoyingly, I suspect their stock will remain just a tad over-priced for the foreseeable future. :-(
Most of this post was essentially correct, but I just wanted to amplify this part of the message. Yes, if you look at the spectral sensitivities of red, green and blue cones (or, strictly, their dyes), blue is many nanometers shorter in wavelength than the difference between red and green. But to test your understanding of how color "works" at the retinal level, the key question should be: where does "yellow" come from?
The answer, of course, is from the additive contributions of both red and green cones; indeed, when you look at the sensitivity curves, you can see that the response to "yellow" should be larger than either green or red. And, it is. Visual acuity is actually slighly better for yellow than for any of the primaries (think shooter's glasses). Now, having said that, I should point out that blue is a special weird case, since the blue cones have a much more limited distribution on the retina than do red and/or green cones.
And, having said all of this, the most amazing thing about color vision (in my opinion) is not what happens at the retina, but what happens in the cortex, apparently in area V4. That's where the very hard problem of color constancy (aka "discounting the illuminant") is solved in a manner studied at great length by Edwin Land, who really would be every geek's hero if only he were better known.
Well, *I* could get the QB to pass, as could my little brother.
The only difference was that his QB always threw incomplete passes, while mine always threw interceptions. Indeed, most of the points scored in our games happened when my QB threw an interception to his "magic guy", if you know what I mean.
On the bright side, the rest of the points in our games happened when his lame ass kicker dribbled a kick on the ground to my "magic guy", or the occasional safety happened. I think I did kick a field goal, once.
Way better than this was the rods-and-slots hockey game, where the "players" were these cardboard things that snapped onto the "stick" bases; you could get different teams in different "uniforms"... On the down side, no magnets were involved. :-)
Others have said this, too, but I'm afraid that I don't know what they're talking about. If you would like to see some careful work on margins of victory (especially in Florida, surf on over to Professor Rusin's paper on this matter. After reading this, you should be clear on the concept that a margin of 200,000 votes out of 100,000,000 cast is highly unlikely to be reversed on any kind of more careful recount unless the original count suffered from some systematic bias affecting one candidate or another. You will then also find out how likely it is that, e.g., the Florida election outcome would be reversed by recounting ballots in counties where you know there to be systematic effects. Basically, Florida would still be close to a statistical dead heat in the situation where we have the totals we have now and the knowledge that there are thousands of untabulated ballots in counties known to prefer one candidate or another.
The questions here are political rather than mathematical, which is why the situation is so awkward for both sides. So not only can we say that recounts (like those we've had in florida) could result in shifting of numbers, but we can also reasonably presume that the slightest change in behavior of either of the candidates could have overcome that margin (i.e., under a popular vote).
So, is this about the University of Missouri Email Wars, or is this Yet Another University Email War (YAUEW!)? I ask, because your whole message sounds eerily familiar. :-)
At Mizzou, we have thousands and thousands of users who have either "missouri.edu" addresses (faculty and staff) or "mizzou.edu" addresses (lowly students). Our system is unstable, slow, and almost certainly not worth it despite an IT squad that is generally smart, responsive, and eager to please. Frankly, I would never have thought I could care less what the email server was that I was using, but now, alas, I know better. :-(