Actually, I do. They use a clear hologram to project light into your eye. The hologram is clear so that you can see through it. It also scatters light into your eye. The light comes from a lens or laser that's mounted on the dashboard or the ceiling or something. The hologram isn't what does the projection -- it just scatters the light back to the pilot.
The important point is that different pixels have to come from different places, so that they could get into the pilot's eye at different angles (so that they'd hit different parts of the pilot's retina). You still can't beat the apparent-size-of-the-last-optic problem, because light travels in straight lines when it's not interacting with optics.
No, not at all. Your LCD projector works by mapping pixels to angle. Each pixel in the image goes out at a different angle, so it hits the screen at a different place.
But the screen is a crucial part of the LCD projector system. If you try to beam images straight into your eye with the projector, what do you see? Try it sometime -- stand in front of the screen. You see a bright lens that looks really tiny, because it's only 3" across and halfway across the room.
The only way to project images directly into your eye is with a lens whose apparent size is larger than the image you want to project (``apparent'' because you can use a close, small lens like a camera viewfinder, or a large, distant lens like those old Fresnel-lens projection televisions...)
If these guys are relying on you to hold their tiny 3" screen up to your head, they've just reinvented viewfinders, which is no big deal. If they want you to project images on the wall, they've just reinvented projectors, also no big deal. They seem to be claiming that they've invented something else entirely -- a screenless projector, if you will, like the interface that Hiro uses in The Diamond Age, but without the cool shades to scatter the light into his eyes. That's not possible, for the reasons I described (apparently not too clearly) above.
No, no, those devices all work on a different principle. They send light into your eye as (more-or-less) collimated beams!
Take your simple Viewmaster. Holding a viewmaster slide right up against your eye illuminates different parts of your pupil with different bits of image -- light passes from the sun or the room lights or whatever through the slide and onto your eye, so there's a little image of the slide projected onto your pupil. What do you see? A blurry mess.
Now stick the Viewmaster slide into the viewer. Lenses in the viewer convert the positional information on the slide into angular information that your eye can process. What do you see? A nice picture of a dinosaur, or whatever.
The point is that the image can only be as big as the apparent size of the lens in the viewmaster. These guys have lots of graphics showing tiny lenses projecting into your eye from far away. That can't work the way that they say. The lens has to be able to get "at" all the different angles coming out of your eye.
It seems to me that they have a sort of (but not very) interesting technology and they're hyping it as the Next Big Thing. But the Big Thinginess comes from applications that are physically impossible. You don't need a laser diode and a scanning mirror to make a ViewMaster work, and there are very nice VR goggles and such that use conventional (if small) LCD displays.
Your eye is a simple camera. The cornea-and-lens assembly on the front has the job of mapping incoming angle of light to particular positions on your retina. That's how you distinguish the angle of incoming light. To generate an image on your retina, you have to change the angle at which light is incident on your eye (provided that the eye is focused properly). Since light travels in straight lines, the only way to do that is to have it arrive from different places outside your eye.
Their little scanning laser thingie can scan a beam across your eye, sure, but if your eye is focused properly the position of the final spot on your retina is independent of where the beam comes into your eye. If the spot position depended on which part of your pupil the beam passed through, then your eye wouldn't be in focus -- normally light from a given object (like the screen you're staring at now) comes through all parts of your pupil simultaneously, so the sharpness of what you see depends critically on your lens getting the job done right. So it doesn't matter how they scan their little mirror-and-laser gismo, they aren't scanning the bright spot on your retina -- they're just shining a blinkenlight at your eye.
And, yes, this argument applies to the cool gizmos in Diamond Age, too. They just don't work.
Now, if you defocus your eye, deliberately NOT looking at the projector gizmo, the system might be able to work. Try it now: hold your thumb right in front of your eye. (Take off your glasses if you have to.) The edge of your thumb looks fuzzy, right? That's because light from the edge of your thumb is passing through several parts of your pupil, and your lens is NOT set correctly to focus that light onto your retina: light from different parts of the pupil hits the retina in different places.
That opens up a nice little loophole: if you deliberately defocus your eye, then the Microvision gizmo could conceivably use that defocus to map position on your pupil to position on the image, and project a nice image on your retina directly. That works in principle, but in practice is neither small nor cheap: they'd have to have some kind of machine vision to track your pupil, at the very least, and that kind of stuff is still expensive.
I wonder if that site is one of those FTC trolls?
Not quantum gravity; semiclassical experiment!
on
Quantum Gravity Observed
·
· Score: 3, Informative
The results described are fabulous, but please don't think that they are sensing quantum gravity in the sense of gravitons -- the postulated gravitational equivalent to photons!
The experiment treats the Earth's gravity well as just another semiclassical potential well. You could get the same effect with protons and a very, very weak electric field (for example).
Not to belittle the experiment -- it's groundbreaking and interesting., and I (for one) can't wait to see a semiclassical quantum verification of the equivalence principle.
It's just not "quantum gravity" in the sense one might naturally think.
Even ``morphological'' studies are no longer done with magnifying
glasses and film, but rather with large collections of digital images from
spacecraft such as SOHO and
TRACE. Image calibration and
reduction software is mandatory if one is to do meaningful
experimental analysis.
Fortunately, the solar community has by-and-large been good about
releasing analysis tools into the public domain -- in fact, there's a
homebrew distribution
system that grew up, mostly before CVS, to nearly-universal status
within the research community. Without the tools that are available
via solarsoft, I literally could not do the work that I do without developing
similar things myself (in fact, I do develop tools myself, and publish
them... but that's another story)
Even within the relatively open solar community, there are
software-based barriers to entry. For example, most of the current
community develops in a proprietary language called IDL, which was developed
in significant part (in its early years) with public funds. The
developer, David Stern, started RSI,
inc. to capitalize on his language. Currently, IDL licenses
start at $1,000 per year, double the current cost of
an entry-level workstation.
When workstations cost $10,000 and only large organizations
could afford hardware capable of doing image processing, this cost was
excusable. But now, in an era of cheap computers, high connectivity,
and readily available space-borne solar data, the cost of supporting
IDL is the main barrier preventing hobbyists, high school students,
and interested amateurs from doing their own research programs.
If IDL were open-source and free, RSI might well still exist (under the
Cygnus / Red-Hat business model), and solar (and other) research would
be much more accessible to the masses.
One may argue that IDL (and its competing product, MatLab) wouldn't
have developed into the large, powerful packages that they are without
commercialization. But such arguments are spurious: PDL, the Perl Data Language, is entirely
open-source and free, and powerful enough that that I am now devloping
tools in it instead of in IDL.
I signed the petition, and I encourage you to, too. Publicly funded
intellectual property is your property, just as the national
forests are your forests. Demand them.
Speakeasy 1.5/384 is no more (128 only on ADSL)
on
Broadband Obstacles
·
· Score: 2
I left Speakeasy in November because they no longer offer any ADSL options with more than 128kb uplink.
I had been using ADSL for a 640k/128k (ISTR) connection and wanted to upgrade to 1.5/384.
The ADSL/SDSL switch is a price point for them now: SDSL costs you significantly more per kbps per month than does ADSL.
I currently use Peak to Peak Internet in Boulder. They're a regional provider and very friendly.
Lunar power stations have day/night cycles... duh.
on
Lunar Lasers
·
· Score: 2
The main problem with solar power systems, at least in places like the Mojave Desert where they've actually been deployed, isn't weather. It's the day/night cycle. You only get about 1/3 of the peak power of the system, on average, even in perfect sunlight. That's because the collecting area is rotating (with the Earth) and almost never has an ideal presentation angle. Then, of course, you get NOTHING during the night, when the Earth is between the collectors and the Sun.
These lunar systems will suffer from the same problem: at most about 1/3 of the peak collecting power will be available on average. Rectennas are pretty cheap compared to solar arrays, but it seems to me that each joule you make on the Moon and beam back to Earth is a pretty expensive one.
Never mind that the beam has to track stations and (to prevent wasting the resource during night from the first receiving station) has to jump between receiving stations that are widely separated in longitude. Lots of opportunities to screw up and irradiate populated areas.
The proprietors say the microwaves are perfectly safe for people -- but the government guidelines for microwave exposure are based on bulk heating effects, not on
any special physics from the waves themselves. That's a bit fishy in itself -- but what about places like hospitals that are filled with sensitive life-support equipment? I can imagine Homer Simpson on the Moon accidentally beaming New York and killing thousands of pacemaker owners and hospital patients.
Passive fiber links let you use IR remote controls
anywhere in the house. A friend of mine has his
stereo (totally IR controlled) in the basement.
He dropped a fiberoptic line from a modified doorbell bracket in the living room, down to
a terminus right in front of the stereo's photocell. Result: aim the remote control at the decorative bracket in the living room, and you can control the invisible stereo. Surprisingly nice to have such an inobtrusive install.
What amazed me is that nonterminated cable is just fine for that purpose! He literally just fished the glass (and cladding) through the wall and mounted it in brackets at each end.
You *can* install drivers in linux without
rebooting. They're called kernel modules.
You can either download an rpm with a collection,
or else compile the exact modules you need
from the kernel source tree. No rebooting required, in most cases.
You do have a good point about ease of install, though. For systems that are in the M$ databases, the installer wizards are pretty slick.
Actually, in the closest analogous medium, the telephone network, it's standard operating procedure. You don't think that the telephone company can simultaneously connect phone calls from everyone in your neighborhood, do you? No, they oversell their network based on the statistics
of residential calling. Dialup internet service (among other things) skews those statistics and requires a lower fanout ratio because many dialup users phone in and stay phoned in for hours at a time. That's why the phone companies want(ed) to charge you extra for modem use.
Business users, who use their phones more, get charged more.
Typical ISPs have fanout ratios of 50 or more for residential service. That's how they can route you for only $50/month or whatever. If you want high speed dedicated service, you have to pay for it. That's all.
The problem is that the cost structure of ISP services doesn't match the
pricing structure. Charging per bit moved wouldn't work, because for
most residential service the main cost is infrastructure support
(the cost of maintaining the pipe, regardless of whether it's used).
But charging only for access, as is currently done, doesn't reflect the
scarcity of the actual resource -- bits moved.
The only reason we (residential customers) have to sign no-resale
agreements is that the ISP's pricing structure is a poor match to the
cost structure. Think about it: if the match were better in the
high-demand case, then no agreement would be necessary. Does the
power company forbid you from reselling your power? No -- but it
doesn't make economic sense for you, because the price structure
matches the cost OK in the high demand case.
The no-redistribution agreeent is a kludge that doesn't even work to
limit customer bandwidth in all cases. Typical ISPs might oversell their pipes by a factor of 50,
so each user must stay below 1/50 of their long-term-average bandwidth or else
the ISP loses money. I just upgraded my DSL connection to 640kb symmetric,
and one use I'm putting the pigger pipe to is listening (at work) to my
home mp3 jukebox. That uses 128kbps, or just over 1/5 of my pipe -- so my
ISP, who charges only for access, loses out on the deal if I leave the stereo
running all day.
A low-volume NATted subnet doesn't affect the fan-out rate nearly
as much as a heavy data mover like my mp3 stream -- though it does use
slightly more bandwidth. A high-volume NATted subnet increases the
spikiness of the load on the ISPs pipe and requires beefier
infrastructure -- so you should pay for it.
It seems to me that the ISPs that charge nothing up to some volume
of data flow, then a fee per gigabyte above that, have the right idea.
That charging scheme matches well with the actual cost of high-volume
users. (Cell-phones work that way too...)
I receive 100% of my salary via federally funded research grants. I find it hard to believe that this issue is anything but clear-cut. The public are paying researchers such as me to find out about the Universe (in this case, about our star, the Sun); if I withheld research results, data, or tools that came from that research, I would be stealing from... you.
If universities withhold software, data, or other knowledge developed using federal research money, they are stealing from... you and me both.
By the way, thanks! Last year, about $0.10 of your tax bill (on average) went to solar physics research.
Just what we need: a computer that's capable of making 20,000,000 mistakes per second, mixed in with 9,980,000,000 right answers.
How do you tell which ones are which?
Speakeasy good -- but not flexible.
on
Geek Gift Ideas 2001
·
· Score: 3, Informative
I used Speakeasy for a while, but eventually ducked out. They have recently taken steps to segment their market: when we joined, they had a large number of ADSL options, but when we went to upgrade our speed (from 608/128 to 608/384 or even 1.0/608), we found that they no longer offer any uplink speed other than 128k, with ADSL. Switching to SDSL means premium pricing and purchase of a new modem.
We ended up going with Peak To Peak DSL -- their service and prices are good (in the Colorado Front Range area). In the Bay Area, I'd recommend CLIQ internet service -- they offer high powered "geek-friendly" DSL.
Don't get me wrong, Speakeasy are good -- but I think they shot themselves in the foot by getting rid of their intermediate-level uplink speeds.
So stop whining already and point your name client at OpenNIC,
the non-ICANN name space. (Of course, OpenNIC
includes the ICANN name space as a subdomain).
It'll just get up and leave because of the
recoil forces on the Francium (formerly Actinium)
nucleus. (A) it won't be bound anymore because the electron orbitals will change, and the chemistry of Americium is different than the chemistry of Francium; and (B) the recoil energy is large compared to the strength of chemical bonds.
Alpha emitters are great for this kind of work,
because alpha particles have a high interaction cross section once they're inside the body. That concentrates their damage in a small space. (You can handle blocks of alpha-decay material without
hazard, because the alpha particles plough into
your epidermis and stop there, wreaking terrible
damage on... tissue that's already dead.)
I bopped on over to one of the online
charts of the nuclides to check out the decay chain of Ac-225.
Indeed, the next two daughters are alpha-emitters, but the first one, Fr-221,
has a 5-minute half-life. That ought to give it plenty of time to get ducted around into your bloodstream and into the rest of your body before emitting the next two alphas and a couple of beta particles, eventually transmuting to stable Bismuth.
So the developers aren't being quite candid when they say that the daugter alpha particles could inflict additional damage on the tumor. Sure, they could -- but (with the antibody bonds long since broken by the recoil from the initial decay) that atom could end up anywhere in your body before decaying again.
This stuff is interesting -- I used to make radioactive saline at the Reed Reactor Facility for medical uses, so I poked around the chart of the nuclides to see how one would make Ac-225. Ideally, you want to start with a nice, stable (or at least long-lived) element, kick a neutron into it (by lowering the ore into a nuclear reactor), and let it turn into what you want via a series of rapid decays. (That's one way to make the Americium 241 in smoke detectors; I'll leave the source element as an exercise for the reader). But Ac-225 doesn't seem to have any such nice precursor decay paths with short half-lives. The half-life is short enough that you wouldn't want to get it from spent fuel (too `hot' until after the Ac-225 is gone!), so I'm not entirely sure how you'd make it.
How quickly we forget...
on
New AIBO Demo'd
·
· Score: 1, Troll
I broke into unix administration in college. I
was working on a fluid-dynamic simulation code
and made the mistake of asking the school's
sysadmins to mount/unmount/dump/restore/whatever
one time too many. Found myself (reluctantly)
with the root password. Then found out that
Next machines (this was 1989; NextOS 0.8beta had
just been released) ran much faster than 1/4 of the school's 11/780. Soon I had root keys to
four NeXT machines running mach, a DECstation 3100, some sort of (then-) screaming fast 386 machine running SVR4,and a sort of klunky Tektronix terminal. The main guidance I got was, "Don't break anything, and whatever you do don't bother us."
That turned out to be excellent experience, and in graduate school I administered a VAX and a couple of Ultrix workstations just because we needed to do work on 'em and there
wasn't any money for a sysadmin.
My point? Just start administering whatever way you can. The problem-solving and man-page-reading skills are far more important than anything you might put on a resume.
Most of the software developed for NASA projects
these days is open -- at least, the scientific
operations and data analysis software. For
example, check out the solarsoft distribution of solar physics analysis software, including planning tools for
most existing solar instruments. CVS and Sourceforge it ain't -- but you can get your hands on the actual software that is being used in the SOHO, TRACE, Yohkoh, and HESSI missions (and soon STEREO and Solar-B too).
I'm delighted to see that someone's reprinted SCMG! Click through to Fatbrain now and buy it!
The book is definitely aimed at children and not adults, but the concepts are interesting and cool enough that it's interesting geek reading too. I still occasionally find myself reciting erudite-yet-silly bits from it, and my original copy is one of my most treasured posessions.
The book is full of translations of "My Black Hen" into a zillion languages -- they're enough to make
any kid think about philosophy:
Probably, Possible, my black hen.
She lays eggs in the relative when.
She doesn't lay eggs in the positive now,
Because she's unable to postulate how!
There's no guarantee that SCMG will make your kid an astrophysicist -- but it'll sure help...
If they don't need broadband, use analog-cell-phone modems. There are cellular-modem PCMCIA cards. With a short ppp timeout, fixed IP, and autodial on net connect, this should work quasi-transparently. The fixed IP is sweet because it allows persistent TCP/IP connections even if you down the link (thereby saving cellular minute charges).
Escrow services like Tradenable (formerly i-escrow) and even Billpoint let you trust the largish corporation (which has much more incentive not to cut and run) instead of the seller. When buying anything over $100, I always use escrow. Sure it costs a couple of bucks more, and there's a tiny amount of added hassle (you have to go back to the web site to verify that the stuff arrived on time), but the security is great and the service really isn't very expensive.
On several occasions, shady-sounding individuals have backed out of auction deals with me, after I suggested using Tradenable. To me, that's the surest sign of a huckster and a good indication that escrow works.
Slashdot Mirror
The important point is that different pixels have to come from different places, so that they could get into the pilot's eye at different angles (so that they'd hit different parts of the pilot's retina). You still can't beat the apparent-size-of-the-last-optic problem, because light travels in straight lines when it's not interacting with optics.
But the screen is a crucial part of the LCD projector system. If you try to beam images straight into your eye with the projector, what do you see? Try it sometime -- stand in front of the screen. You see a bright lens that looks really tiny, because it's only 3" across and halfway across the room.
The only way to project images directly into your eye is with a lens whose apparent size is larger than the image you want to project (``apparent'' because you can use a close, small lens like a camera viewfinder, or a large, distant lens like those old Fresnel-lens projection televisions...)
If these guys are relying on you to hold their tiny 3" screen up to your head, they've just reinvented viewfinders, which is no big deal. If they want you to project images on the wall, they've just reinvented projectors, also no big deal. They seem to be claiming that they've invented something else entirely -- a screenless projector, if you will, like the interface that Hiro uses in The Diamond Age, but without the cool shades to scatter the light into his eyes. That's not possible, for the reasons I described (apparently not too clearly) above.
Take your simple Viewmaster. Holding a viewmaster slide right up against your eye illuminates different parts of your pupil with different bits of image -- light passes from the sun or the room lights or whatever through the slide and onto your eye, so there's a little image of the slide projected onto your pupil. What do you see? A blurry mess.
Now stick the Viewmaster slide into the viewer. Lenses in the viewer convert the positional information on the slide into angular information that your eye can process. What do you see? A nice picture of a dinosaur, or whatever.
The point is that the image can only be as big as the apparent size of the lens in the viewmaster. These guys have lots of graphics showing tiny lenses projecting into your eye from far away. That can't work the way that they say. The lens has to be able to get "at" all the different angles coming out of your eye.
It seems to me that they have a sort of (but not very) interesting technology and they're hyping it as the Next Big Thing. But the Big Thinginess comes from applications that are physically impossible. You don't need a laser diode and a scanning mirror to make a ViewMaster work, and there are very nice VR goggles and such that use conventional (if small) LCD displays.
Their little scanning laser thingie can scan a beam across your eye, sure, but if your eye is focused properly the position of the final spot on your retina is independent of where the beam comes into your eye. If the spot position depended on which part of your pupil the beam passed through, then your eye wouldn't be in focus -- normally light from a given object (like the screen you're staring at now) comes through all parts of your pupil simultaneously, so the sharpness of what you see depends critically on your lens getting the job done right. So it doesn't matter how they scan their little mirror-and-laser gismo, they aren't scanning the bright spot on your retina -- they're just shining a blinkenlight at your eye.
And, yes, this argument applies to the cool gizmos in Diamond Age, too. They just don't work.
Now, if you defocus your eye, deliberately NOT looking at the projector gizmo, the system might be able to work. Try it now: hold your thumb right in front of your eye. (Take off your glasses if you have to.) The edge of your thumb looks fuzzy, right? That's because light from the edge of your thumb is passing through several parts of your pupil, and your lens is NOT set correctly to focus that light onto your retina: light from different parts of the pupil hits the retina in different places.
That opens up a nice little loophole: if you deliberately defocus your eye, then the Microvision gizmo could conceivably use that defocus to map position on your pupil to position on the image, and project a nice image on your retina directly. That works in principle, but in practice is neither small nor cheap: they'd have to have some kind of machine vision to track your pupil, at the very least, and that kind of stuff is still expensive.
I wonder if that site is one of those FTC trolls?
The experiment treats the Earth's gravity well as just another semiclassical potential well. You could get the same effect with protons and a very, very weak electric field (for example).
Not to belittle the experiment -- it's groundbreaking and interesting., and I (for one) can't wait to see a semiclassical quantum verification of the equivalence principle.
It's just not "quantum gravity" in the sense one might naturally think.
Even ``morphological'' studies are no longer done with magnifying glasses and film, but rather with large collections of digital images from spacecraft such as SOHO and TRACE. Image calibration and reduction software is mandatory if one is to do meaningful experimental analysis.
Fortunately, the solar community has by-and-large been good about releasing analysis tools into the public domain -- in fact, there's a homebrew distribution system that grew up, mostly before CVS, to nearly-universal status within the research community. Without the tools that are available via solarsoft, I literally could not do the work that I do without developing similar things myself (in fact, I do develop tools myself, and publish them... but that's another story)
Even within the relatively open solar community, there are software-based barriers to entry. For example, most of the current community develops in a proprietary language called IDL, which was developed in significant part (in its early years) with public funds. The developer, David Stern, started RSI, inc. to capitalize on his language. Currently, IDL licenses start at $1,000 per year, double the current cost of an entry-level workstation.
When workstations cost $10,000 and only large organizations could afford hardware capable of doing image processing, this cost was excusable. But now, in an era of cheap computers, high connectivity, and readily available space-borne solar data, the cost of supporting IDL is the main barrier preventing hobbyists, high school students, and interested amateurs from doing their own research programs. If IDL were open-source and free, RSI might well still exist (under the Cygnus / Red-Hat business model), and solar (and other) research would be much more accessible to the masses.
One may argue that IDL (and its competing product, MatLab) wouldn't have developed into the large, powerful packages that they are without commercialization. But such arguments are spurious: PDL, the Perl Data Language, is entirely open-source and free, and powerful enough that that I am now devloping tools in it instead of in IDL.
I signed the petition, and I encourage you to, too. Publicly funded intellectual property is your property, just as the national forests are your forests. Demand them.
I left Speakeasy in November because they no longer offer any ADSL options with more than 128kb uplink.
I had been using ADSL for a 640k/128k (ISTR) connection and wanted to upgrade to 1.5/384.
The ADSL/SDSL switch is a price point for them now: SDSL costs you significantly more per kbps per month than does ADSL.
I currently use Peak to Peak Internet in Boulder. They're a regional provider and very friendly.
These lunar systems will suffer from the same problem: at most about 1/3 of the peak collecting power will be available on average. Rectennas are pretty cheap compared to solar arrays, but it seems to me that each joule you make on the Moon and beam back to Earth is a pretty expensive one.
Never mind that the beam has to track stations and (to prevent wasting the resource during night from the first receiving station) has to jump between receiving stations that are widely separated in longitude. Lots of opportunities to screw up and irradiate populated areas.
The proprietors say the microwaves are perfectly safe for people -- but the government guidelines for microwave exposure are based on bulk heating effects, not on any special physics from the waves themselves. That's a bit fishy in itself -- but what about places like hospitals that are filled with sensitive life-support equipment? I can imagine Homer Simpson on the Moon accidentally beaming New York and killing thousands of pacemaker owners and hospital patients.
What amazed me is that nonterminated cable is just fine for that purpose! He literally just fished the glass (and cladding) through the wall and mounted it in brackets at each end.
The render is REALLY MANGLED with Opera 5. What kind of dumbass company helps clients harness the Internet'' but can't even write web-standards-compliant html?
You *can* install drivers in linux without rebooting. They're called kernel modules. You can either download an rpm with a collection, or else compile the exact modules you need from the kernel source tree. No rebooting required, in most cases.
You do have a good point about ease of install, though. For systems that are in the M$ databases, the installer wizards are pretty slick.
>In other venues, it's called fraud.
Actually, in the closest analogous medium, the telephone network, it's standard operating procedure. You don't think that the telephone company can simultaneously connect phone calls from everyone in your neighborhood, do you? No, they oversell their network based on the statistics of residential calling. Dialup internet service (among other things) skews those statistics and requires a lower fanout ratio because many dialup users phone in and stay phoned in for hours at a time. That's why the phone companies want(ed) to charge you extra for modem use.
Business users, who use their phones more, get charged more.
Typical ISPs have fanout ratios of 50 or more for residential service. That's how they can route you for only $50/month or whatever. If you want high speed dedicated service, you have to pay for it. That's all.
The problem is that the cost structure of ISP services doesn't match the pricing structure. Charging per bit moved wouldn't work, because for most residential service the main cost is infrastructure support (the cost of maintaining the pipe, regardless of whether it's used). But charging only for access, as is currently done, doesn't reflect the scarcity of the actual resource -- bits moved.
The only reason we (residential customers) have to sign no-resale agreements is that the ISP's pricing structure is a poor match to the cost structure. Think about it: if the match were better in the high-demand case, then no agreement would be necessary. Does the power company forbid you from reselling your power? No -- but it doesn't make economic sense for you, because the price structure matches the cost OK in the high demand case.
The no-redistribution agreeent is a kludge that doesn't even work to limit customer bandwidth in all cases. Typical ISPs might oversell their pipes by a factor of 50, so each user must stay below 1/50 of their long-term-average bandwidth or else the ISP loses money. I just upgraded my DSL connection to 640kb symmetric, and one use I'm putting the pigger pipe to is listening (at work) to my home mp3 jukebox. That uses 128kbps, or just over 1/5 of my pipe -- so my ISP, who charges only for access, loses out on the deal if I leave the stereo running all day.
A low-volume NATted subnet doesn't affect the fan-out rate nearly as much as a heavy data mover like my mp3 stream -- though it does use slightly more bandwidth. A high-volume NATted subnet increases the spikiness of the load on the ISPs pipe and requires beefier infrastructure -- so you should pay for it.
It seems to me that the ISPs that charge nothing up to some volume of data flow, then a fee per gigabyte above that, have the right idea. That charging scheme matches well with the actual cost of high-volume users. (Cell-phones work that way too...)
By the way, thanks! Last year, about $0.10 of your tax bill (on average) went to solar physics research.
How do you tell which ones are which?
We ended up going with Peak To Peak DSL -- their service and prices are good (in the Colorado Front Range area). In the Bay Area, I'd recommend CLIQ internet service -- they offer high powered "geek-friendly" DSL.
Don't get me wrong, Speakeasy are good -- but I think they shot themselves in the foot by getting rid of their intermediate-level uplink speeds.
So stop whining already and point your name client at OpenNIC, the non-ICANN name space. (Of course, OpenNIC includes the ICANN name space as a subdomain).
It'll just get up and leave because of the recoil forces on the Francium (formerly Actinium) nucleus. (A) it won't be bound anymore because the electron orbitals will change, and the chemistry of Americium is different than the chemistry of Francium; and (B) the recoil energy is large compared to the strength of chemical bonds.
Alpha emitters are great for this kind of work, because alpha particles have a high interaction cross section once they're inside the body. That concentrates their damage in a small space. (You can handle blocks of alpha-decay material without hazard, because the alpha particles plough into your epidermis and stop there, wreaking terrible damage on ... tissue that's already dead.)
I bopped on over to one of the online charts of the nuclides to check out the decay chain of Ac-225. Indeed, the next two daughters are alpha-emitters, but the first one, Fr-221, has a 5-minute half-life. That ought to give it plenty of time to get ducted around into your bloodstream and into the rest of your body before emitting the next two alphas and a couple of beta particles, eventually transmuting to stable Bismuth.
So the developers aren't being quite candid when they say that the daugter alpha particles could inflict additional damage on the tumor. Sure, they could -- but (with the antibody bonds long since broken by the recoil from the initial decay) that atom could end up anywhere in your body before decaying again.
This stuff is interesting -- I used to make radioactive saline at the Reed Reactor Facility for medical uses, so I poked around the chart of the nuclides to see how one would make Ac-225. Ideally, you want to start with a nice, stable (or at least long-lived) element, kick a neutron into it (by lowering the ore into a nuclear reactor), and let it turn into what you want via a series of rapid decays. (That's one way to make the Americium 241 in smoke detectors; I'll leave the source element as an exercise for the reader). But Ac-225 doesn't seem to have any such nice precursor decay paths with short half-lives. The half-life is short enough that you wouldn't want to get it from spent fuel (too `hot' until after the Ac-225 is gone!), so I'm not entirely sure how you'd make it.
I'd never buy an AIBO. Sony blew it. I'd rather build my own robot out of
Lego Mindstorms stuff.
I broke into unix administration in college. I was working on a fluid-dynamic simulation code and made the mistake of asking the school's sysadmins to mount/unmount/dump/restore/whatever one time too many. Found myself (reluctantly) with the root password. Then found out that Next machines (this was 1989; NextOS 0.8beta had just been released) ran much faster than 1/4 of the school's 11/780. Soon I had root keys to four NeXT machines running mach, a DECstation 3100, some sort of (then-) screaming fast 386 machine running SVR4,and a sort of klunky Tektronix terminal. The main guidance I got was, "Don't break anything, and whatever you do don't bother us."
That turned out to be excellent experience, and in graduate school I administered a VAX and a couple of Ultrix workstations just because we needed to do work on 'em and there wasn't any money for a sysadmin.
My point? Just start administering whatever way you can. The problem-solving and man-page-reading skills are far more important than anything you might put on a resume.
Most of the software developed for NASA projects these days is open -- at least, the scientific operations and data analysis software. For example, check out the solarsoft distribution of solar physics analysis software, including planning tools for most existing solar instruments. CVS and Sourceforge it ain't -- but you can get your hands on the actual software that is being used in the SOHO, TRACE, Yohkoh, and HESSI missions (and soon STEREO and Solar-B too).
The book is definitely aimed at children and not adults, but the concepts are interesting and cool enough that it's interesting geek reading too. I still occasionally find myself reciting erudite-yet-silly bits from it, and my original copy is one of my most treasured posessions.
The book is full of translations of "My Black Hen" into a zillion languages -- they're enough to make any kid think about philosophy:
Probably, Possible, my black hen.
She lays eggs in the relative when.
She doesn't lay eggs in the positive now,
Because she's unable to postulate how!
There's no guarantee that SCMG will make your kid an astrophysicist -- but it'll sure help...
If they don't need broadband, use analog-cell-phone modems. There are cellular-modem PCMCIA cards. With a short ppp timeout, fixed IP, and autodial on net connect, this should work quasi-transparently. The fixed IP is sweet because it allows persistent TCP/IP connections even if you down the link (thereby saving cellular minute charges).
On several occasions, shady-sounding individuals have backed out of auction deals with me, after I suggested using Tradenable. To me, that's the surest sign of a huckster and a good indication that escrow works.