I’ve been working on GPSD a lot recently – we’re heading towards a 3.10 release with a lot of new features. As part of this release I’ve decided to ship a HOWTO on setting up a high-quality NTP time server using GPSD. In the course of working on that, I’ve had an idea.
The idea has two antecedents. One is that if you start with any one of several inexpensive GPS modules (my favorite of which is the u-blox 6), and add GPSD to read it and feed an ntpd instance, it’s possible to build an NTP server that meets the usual standard for public Stratum 1 time servers – 10mSec or better accuracy to UTC.
The other is that there is a raft of inexpensive SBCs that run Linux out there – Arduino, Raspberry Pi and the current new hotness BeagleBone. So here’s my thought: why not build a low-power Stratum 1 timeserver on a credit-card-sized SBC?
A little googling tells me that someone has already mated a GPS to a BeagleBone Black. His design uses a u-blox, but has one serious drawback – he doesn’t wire up 1PPS from the receiver. This is a must if you want sub-second precision on your time service.
Now, to be fair, the D2523T he’s using might not bring 1PPS out from the u-blox5 to the module header. On the other hand, I looked up the D2523T’s data sheet on Sparkfun and their pinout for it has a lead labeled “GPS LED”.
Bingo! Boys and girls, the LED on your GPS is driven by 1PPS – the once-per-second flash it emits is top of GPS second accurate to 50 nanoseconds. It seems likely that we could take the right small variant of this design, add ntpd to it, do a little configuration, and stand back! Instant Stratum 1 in a smaller package than your smartphone, powered off its USB cable.
Well, that is, if the RS232 headers on the BeagleBone include any handshake line. Carrier Detect, Ring Indicator, even Clear To Send – any of these would be good enough. 1PPS has to travel over something that isn’t TX.
It would be a fun project. I wish I had time to do it.
The P8 expansion header on the BeagleBone lists CTS lines for UARTs 3, 4 and 5. You’ll still need to translate the LED flash to RS-232, though…wire an optoisolator input in place of the LED?
>The P8 expansion header on the BeagleBone lists CTS lines for UARTs 3, 4 and 5.
Good enough. GPSD can handle 1PPS over CTS.
>You’ll still need to translate the LED flash to RS-232, though…
Nah. On the module, the “GPS LED” is a lead, not an actual LED. At least if I’m reading the datasheet correctly.
“All in a package smaller than your smartphone”. Interesting, I wonder if there are any smartphones that have a GPS that has PPS set up. No hardware hacking required…
Have you looked at the “Laureline” embedded GPS NTP server project? I suspect it’s basically exactly what you are looking for, with the bonus of being both completely open-source, and cheap ($64)!
Design log:
http://www.eevblog.com/forum/oshw/'laureline'-embedded-gps-ntp-server/
On sale on tindie:
https://www.tindie.com/products/gxti/laureline-gps-ntp-server/
>Have you looked at the “Laureline” embedded GPS NTP server project?
Damn, that is sweet! Can’t beat the price, either.
SBC devices usually have craptons of GPIO pins, so there’s no need to emulate one by abusing UART pins. You can configure any GPIO pin to wake up userspace with a microsecond’s latency or even less by echoing a few integers into the /sys/class/gpio hierarchy. You don’t even have to rebuild the kernel to get it.
There are UART CTS lines on the BBB headers, but they’re not exposed by default–with good reason, since you have to sacrifice one of the MMC/SD storage devices to get them (they use the same physical pin). Most things in the SBC embedded world that use UART don’t use flow control (they’re typically very slow, like sensors that don’t generate much data, or very fast, like LCD displays that can receive data faster than it can possibly be sent). The devices that do need flow control (like embedded cellular modems) are usually much easier to use through the BBB’s USB port (with a USB-to-RS232 cable that would include the required voltage level shifters) than through the SoC UART ports. CTS, RTS, DTR, and the other modem control lines are as out of favor today as the modems they used to connect to.
If that PPS output pin really is intended for a LED, it may include a series resistor that will delay detection of the rising edge by a few hundred ns. Still good enough for 0.1 microsecond accuracy though.
I have the Adafruit Ultimate GPS breakout (http://www.adafruit.com/products/790) which is designed to be used with devices like the BBB, Pi, and Arduino. I don’t (yet) use it as a time server (too many projects in my queue…maybe in 2014?).
In case you are interested, I’ve been writing up my experiments with a Raspberry Pi and various GPS modules here:
http://www.satsignal.eu/ntp/Raspberry-Pi-NTP.html
I’ve also used the Adafruit module, and there is a new module coming out soon which is a direct plug-in for the Raspberry pi 26-pin header, and uses the new ublox Max 7Q timing GPS module. No wiring required.
David Taylor runs some Linux RasPi Stratum 1 servers with uBlox, Sure, and Adafruit GPS modules, and a Trimble Resolution T at http://satsignal.eu/mrtg/daily_ntp.html. His site displays the stats and documents all the setup required, including wiring and sources for NTP mods. Others have documented similar setups including the Beaglebone and LEA 6T on comp.protocols.time.NTP and comp.time.nuts groups and mailing lists.
I have an adafruit ultimate breakout mated to a raspberri pi as a stratum 1. There is a custom rasbian kernel on an sd image out there that makes this plug and play… Burn the card…hook up 5 wires (tx,rx,power, gnd, pps) to the pi and it is a stratum 1. There several how-to blogs for this.
The sparkfun Skytraq module connects to the header on the raspberry pi, and there is already a gpio pps userland driver. And I have AGPS I had it going (until it spilled out of my tour pack onto I90 and got run over). But all the parts and code are there, the only problem is you want a different GPIO pin then the recommended 8 since it seems to pull it up or down into a bootup mode. I think I used pin 23.
Something I want to note: the Arduino emphatically does not run Linux.
Also, yeah, throw away the handshaking lines (unless they’re especially convinient). Just grab a spare GPIO pin. There are many of them, which gives you more options to avoid other useful peripherals
>Something I want to note: the Arduino emphatically does not run Linux.
I checked before I wrote that. There are very recent Arduino boards that boot Linux. Mind you, from the hardware specs on them the amazing thing is that the bear dances at all.
many people recommended you do exactly this back when you had the ill-founded idea for PPS over USB.
>many people recommended you do exactly this back when you had the ill-founded idea for PPS over USB.
My “ill-founded idea” turned into an inexpensive and excellent time source, delivering 1ms accuracy (not good enough for Stratum 1 but 10 times better than typical Stratum 2) for $33 a pop. Nothing even remotely as good is available for a price less than an order of magnitude higher. We could use a lot more such ill-foundedness, I think.
While we are back on gpsd, whatever happened to your project to mount and antenna and get a proper gps sat feed into the workroom?
IIRC, the last report had you looking at an antenna + repeater setup, with the antenna mounted on the gable end.
>While we are back on gpsd, whatever happened to your project to mount and antenna and get a proper gps sat feed into the workroom?
The electrician who was going to mount the antenna mast flaked out on me, which annoyed me until I learned (by chance) that he had died. Ill health will do that.
I intend to reschedule the job with someone else, but it’s not likely to happen until the spring. People who will go up ladders in Pennsylvania in cold months are expensive.
Huh? For $330 one could build a complete PPS-reading stand-alone GPS-to-NTP server, with a weatherproof enclosure, WiFi, a solar panel, and a battery so it can live outside where the satellites are. Prices have been coming down, but not nearly that quickly.
Arduino makes a number of products. The one most people know of as “Arduino” is an Atmel AVR microcontroller with up to 2K of RAM and 32K of flash. The AVR doesn’t run Linux. Some of the older/smaller Arduino models aren’t even big enough to hold a PC boot sector in their RAM. The AVR can do things Linux machines can’t, like fit in a cavity designed for an AA battery while using another one as power, or execute code with deterministic instruction timing. 90% of an Arduino’s price is the board with its expansion connectors and USB-to-serial interface–the $4 AVR chip can be removed once programmed and will work all by itself, which makes it popular for projects where you need small amounts of disposable computing, like rocket telemetry.
Arduino (the company) does make Arduino-branded devices that have ARM processors in them which can run Linux, like the YÚN. They’re competitive responses to the flood of low-cost Linux ARM boards coming onto the market like the Beagle Bone Black (which was in turn a response to the Raspberry Pi). Arduino has a bit of a problem here as it’s currently cheaper to buy an AVR chip, USB WiFi, and a Raspberry Pi and wire them together than it is to buy a YÚN, but lots more work. As soon as someone fixes that (i.e. build a cheap PCB with an AVR socket, and push the software fixes to Raspbian or Debian), the YÚN doesn’t make a lot of sense at its current price.
Ironically, arduino (an IDE written in Java, a third distinct thing that shares the name “Arduino”) will run on a Raspberry Pi, so you can connect a Raspberry Pi’s SPI port to an AVR chip and program it with no other hardware; however, avrdude (the program loader) doesn’t have a device driver for Linux SPI, so the arduino IDE doesn’t know how to do this. There are lots of blog posts about this, but little or no upstream integration yet–and upstream might be reluctant to comply as they tend to see Linux SBCs as competitors (Arduino fans complain that Raspberry Pi is eating their lunch). avrdude does know how to do SPI over the parallel printer port–an interface I haven’t seen on a new PC in over a decade, and utterly unhelpful on an ARM.
I presume gpsd has some similar awkwardness when talking to directly-wired GPS receivers, given how far behind the curve ESR seems to be on these SBC devices.
Arduino (the AVR chip) comes with an enormous, diverse software community that produces driver code for things people want to control from AVR chips. It’s a treasure-trove of open-source (or at least source-readable) device drivers that run on bare metal hardware. These days if I pick up some random digital sensor component, chances are medium-to-good that it has both a data sheet and an Arduino sketch online somewhere. There’s an Arduino GPS receiver library which should be an interesting read for anyone who hacks GPS daemons. ;)
>Huh? For $330 one could build a complete PPS-reading stand-alone GPS-to-NTP server, with a weatherproof enclosure, WiFi, a solar panel, and a battery so it can live outside where the satellites are. Prices have been coming down, but not nearly that quickly.
I have no doubt of it. But the key word is “build”. The GR601-W is stable production hardware that you can buy by the hundred, not a custom or semi-custom build sweated by hand in someone’s garage and changing every week. Its only real competitor is a USB timing GPS made by a small company in Brittany (of all places) which is $900 a unit in the U.S. if you can get it at all. (I suspect theirs uses the same single-wire “Macx-1” mod as the GR-601W to the bog-standard two-chips-on-a-miniboard design, which makes that price highway robbery.)
I love concepts like the Laureline and Dave Taylor’s designs. But I don’t look at them the way a hobbyist does; I have an actual volume deployment case in mind. I designed the GR-601W so we could plug 500 of them of them into Internet routers and do time-delay tomography, sanity-checking NTP itself. You can’t practically speaking meet those numbers with a Raspberry Pi or Beaglebone-based hand-build, let alone a full custom design; you need something the worker bees in Shenzhen can crank out for you.
>I presume gpsd has some similar awkwardness when talking to directly-wired GPS receivers, given how far behind the curve ESR seems to be on these SBC devices.
Nope. Looks to me like most of the SBC builds are using GPSD – certainly all the Linux-based ones seem to be. Well, what else could they use? GPS reporting protocols aren’t hard, exactly, but they’re very messy and fiddly. GPSD has no real competition, because nobody else wants to get into that swamp. And if you knew what I do you wouldn’t blame them a bit for staying out.
All GPSD cares about is that the device path you pass it opens as a serial device producing packets in one of the sentence formats GPSD knows about – which is basically every GPS protocol in the known universe plus RTCM plus Marine AIS plus some odds and ends like magnetic compasses and fish sounders. “Directly wired” or any other implementation detail doesn’t really signify.
esr > My “ill-founded idea” turned into an inexpensive and excellent time source, delivering 1ms accuracy (not good enough for Stratum 1 but 10 times better than typical Stratum 2) for $33 a pop.
Yes, I’ve been meaning to get my hand on one of these and play with it. But I couldn’t find what name it trades under and where I can buy it. Can you say it again, please?
>Yes, I’ve been meaning to get my hand on one of these and play with it. But I couldn’t find what name it trades under and where I can buy it. Can you say it again, please?
It’s the Navisys GR601-W. Navisys wants to sell them qty 100, which is fine for my deployment case but a bit awkward for individuals other than me. Email gem@rellim.com; he bought a block of them, I think he wants $45 with U.S shipping included.
> I think he wants $45 with U.S shipping included.
Unit Price USD28.5/pc FOB China.
and… they don’t work without egregious hacks in the kernel. USB imposes too much latency in any case.
You can make a Stratum 1 NTP server out of an Alix and a disciplined oscillator.
The father of one of my employees was the principal engineer for Symmetricom.
>and… they don’t work without egregious hacks in the kernel.
What egregious hacks do you believe are required, and why?
>USB imposes too much latency in any case.
The engineers in Shenzhen measured latency (using GPSD) before they shipped me a sample. In no case did they observe latency significantly greater than the 1ms polling interval. Average latency should be the half-interval time, or about 500μs. I confirmed these results here. So have others.
Maybe you think 1ms is “too much latency” but that is a curious position to take given that Stratum 2 expected accuracy is a full two orders of magnitude worse. 1ms is certainly good enough for the GR601-W’s intended application in WAN delay tomography, where the feature sizes tend to be upwards of 10ms.
If you know a different way to get two orders of magnitude better than Stratum 2 from a $28 device you can hotplug into a USB port, please speak right up. Otherwise, cease thy trolling. You’re beginning to bore me.
Aren’t Stratum 1 devices directly connected to the atomic clocks, with peer-to-peer cross-checking with other clocks?
>Aren’t Stratum 1 devices directly connected to the atomic clocks, with peer-to-peer cross-checking with other clocks?
Yes, if you count a GPS as an atomic clock :-)
Seriously, a lot of Stratum 1s use GPSes – not consumer-grade stuff but timing devices with 1PPS and stationary-mode operation where you sacrifice location capability to get better clock. This is completely reasonable: GPS top-of-second is UTC accurate to within 50ns, modulo some glitches near leap seconds. Even with ionospheric delays and various latency sources in host systems you can hold a GPS-conditioned clock to within 1μs of UTC if you know what you’re doing.
And yes, a fair number of Stratum 1 servers use GPSD to pass time and 1PPS conditioning to NTP. It has little competition for that role, and no effective competitors under Unix that aren’t one-off builds tied to specific custom hardware.
esr > It’s the Navisys GR601-W.
Thanks!
@TomA
> Aren’t Stratum 1 devices directly connected to the atomic clocks, with peer-to-peer cross-checking with other clocks?
Isn’t that what GPSD does?
Sorry, I know I’m overworking this. But if the clock is in orbit, then you have transmission latency and relativistic effects that make it characteristically different from a clock and server in the same cabinet on a fixed spot on the earth’s surface. Maybe there should be Stratum 1a and Stratum 1b.
“Aren’t Stratum 1 devices directly connected to the atomic clocks, with peer-to-peer cross-checking with other clocks?”
Now you run into the old problem of, “two differing measurements – which is right?”. At the TV network that I worked at, we had our own cesium standards to provide our reference timing. If our master clocks were not periodically corrected (first with a modem connection to the Naval Observatory, later with a GPS system), they would drift off by a tenth of a second or so after a few weeks.
I would suggest that everyone just take the attitude that, “GPS has better hardware than anything we can build.” and “GPS is universally available and constantly maintained.” and just accept it as The Standard.
> good enough for the GR601-W’s intended application in WAN delay tomography, where the feature sizes tend to be upwards of 10ms.
It’s probably a good fit for that application, but for a generalized NTP server, it’s gonna suck.
Latency isn’t the issue, jitter (variance of the latency) most likely is.
>Latency isn’t the issue, jitter (variance of the latency) most likely is.
You’re right, but we get a nice serendipity here. USB polling intervals are pretty stable. The time out of a GR601-W is actually way cleaner than I was really expecting from consumer-grade parts. It helps that the u-blox 6 is pretty rock solid as a GPS/1PPS engine – as many experimenters have recently found, even the variants that aren’t officially specialized for time service are impressively good at it.
Droid48 is an android emulation of an HP-48GX. It uses an image of the HP48 ROM.
A real HP48GX has a serial port and the correct cable will allow one to connect to the HP48 with kermit etc.
Does anyone know how the GPS in (for example) a Google Nexus 4 (LG version running Android 4.3 exposes the GPS data stream?
And does anyone know how that stream might be directed to the USB port of an android phone, to be fed to a real HP48, or even better to the internal virtual input of the Droid48 instance. (I believe that the emulation works.)
Worker bees in the US will turn your EDA files into a PCB in your snailmail box for $10 plus shipping. Then you have an interface kit between an existing GPS module design and an existing SBC design. The kit requires only component placement and soldering to assemble, and after building a few prototypes to make sure they work, you can email the EDA files to Shenzhen and order 1000 copies. If the SBC gets discontinued, you just edit yourself a new PCB to fit whatever the new-hawtness SBC is, and keep going (hopefully you’re building in small, demand-driven batches). Chips and modules come and go all the time. Supply chain management is half the fun. ;)
Your $30 USB device adds GPS to a router, and it’s not a bad price given the cost of the individual components; however, for the highway robbery price, or even within an order of magnitude of your price, you could have picked a GPS, and then added a cheap router to it. Or even a non-cheap one.
I guess we’re just violently agreeing that $900 for a USB-GPS module is highway robbery.
At the top of this post you were asking for a UART CTS line, which is an unreasonable request for the SBC world. I admit I was a bit blinded by the WTF–gpsd itself can be happily pointed at any UART for position data, and just work.
I don’t really understand why gpsd wants the PPS signal at all–surely you’d want that routed directly to ntpd? And if so, doesn’t having the PPS signal on CTS actually make that harder?
>At the top of this post you were asking for a UART CTS line
Oh, right, I grasp the source of confusion now.
GPSD doesn’t use 1PPS for anything it does itself. It will happily chew on time/location data from an RS232 interface that only consists of Tx, Rx, and ground. The problem is that the in-band time information from a GPS is both low precision and really wobbly. One guy I discussed measurements with found that the once-per-second start of reporting on a SiRF (the most common consumer-grade GPS) wandered in a sawtooth patten with an amplitude of about 170ms and a period of hours.
1PPS, on the other hand, is within 50ns of GPS top-of-second. What GPSD wants 1PPS for is not its own use but so that it can pass high-accuracy time information to ntpd via a shared-memory segment. For that it needs to see one line carrying 1PPS that is not Tx. It watches for transitions on CD, RI, and CTS.
>I don’t really understand why gpsd wants the PPS signal at all–surely you’d want that routed directly to ntpd? And if so, doesn’t having the PPS signal on CTS actually make that harder?
I see you don’t grok how ntpd works. ntpd doesn’t watch or manage devices itself, ever. It doesn’t know 1PPS from a hole in the ground. What ntpd does is watch a shared-memory segment where symbiont programs (clock or sensor managers) drop timing information in a known format. This is proper separation of function; the work of collecting the data (the sensor’s job) is decoupled from the work of aggregating, analyzing and redistributing it (ntpd’s job).
gpsd, the service daemon that is the core of GPSD, is such a symbiont. It combines in-band GPS data with the 1PPS pulse and uses that to ship high-quality time to ntpd. But, as noted, this can’t happen unless the 1PPS presents itself on something that looks to gpsd like an RS232 handshake pin.
Which is why “Oh, well, you can get 1PPS on a random GPIO pin” is not really a helpful response. Wiring custom cables sucks when you’re trying to produce at volume.
I had seen code that does that, but until now I didn’t know why. I’ve seen device driver code for ntpd in the past…but maybe it’s just vestigial code from the 1990’s?
Right, but random GPIO pins are set up that way on Linux. You can configure them through a sysfs interface, and put their file descriptors into poll() or select() arguments. Or you can set up your own IRQ harvesting kernel module, and snag a high-precision timestamp in the IRQ handler entirely outside of gpsd (someone has a HOWTO for this on the Raspberry Pi). Both of these approaches are things people are already using for NTP+GPS.
Right, but in the original article you didn’t want to make cables–you wanted to make expansion boards that mate with an existing SBC, or modify an open-source SBC design so the GPS expansion is built in. In that case you can pick whatever pins you like, and make the boards at whatever volumes you like. You’re insisting that users give up a precious CTS pin (something that Raspberry Pi users don’t have to give in the first place), when you could be using any random GPIO pin that nobody cares about.
>I’ve seen device driver code for ntpd in the past…but maybe it’s just vestigial code from the 1990?s
Possibly older than that. The present ntpd codebase is descended from software that ran back in the 1980s on specialized PDP-11 timeservers called “fuzzballs”. It’s probable that ancient ntpd was tied to that hardware and some of the driver code for it wasn’t cleaned out until much later.
> You’re insisting that users give up a precious CTS pin (something that Raspberry Pi users don’t have to give in the first place), when you could be using any random GPIO pin that nobody cares about.
Whoa. Sounds to me like you’ve way overinterpreted something I or someone else said.
For one thing, I’m sure I never fixated on CTS, because the traditional pin for this is actually CD (with RI as a rare alternative). CTS is sort of theoretically possible (and gpsd watches it for that reason) but I’ve never heard of it being used for 1PPS in practice.
For a second, I don’t really know enough about hardware engineering to have ambitions as specific as “make expansion boards”. The language you are using to argue against what you think I said is alien to me. Part of the reason I was wistful about doing this project was as a way to learn more.
But really, knowing that things like the Laureline exist and are moving towards maturity is good enough for now.
…and effectively BeagleBone Black doesn’t have it either, since the CTS and RTS pins from one of the UARTs are used for the MMC device instead (possibly the same device where / is mounted). The white BeagleBone white didn’t have any hardware on those pins, so they might have been usable, but I don’t think BB whites are in production any more.
Pinmux is a harsh mistress. Interface pins with specific functions, like UART control lines, are precious and rare and shouldn’t be wasted whenever it’s possible to use a general-purpose pin instead.
They’re all equally bad. There are a ridiculous number of UARTs in a BBB (3? 4?), but only the TX and RX pins are usable. Every other pin either didn’t make it to the header, or it’s only on the header because it happens to share a pin with something else.
Pins on most SoC devices have multiple mutually exclusive functions. Usually they are GPIO and something else, but sometimes one pin gets half a dozen possible modes. It lets designers trade off between e.g. having a 24-bit LCD display, vs a 16-bit LCD display with something more interesting connected to the other 8 pins. Sometimes this is a really useful feature–you can often read a multifunction pin through GPIO even while it’s being used for something else, so you can see what your circuit is doing from the CPU without physically probing it.
Somebody get this man an Arduino Experimentation Kit, stat! ;)
The AVR is a simple enough processor that you can read the entire reference manual, but it’s complex enough that it lets you experience most of the design trade-offs you have to make when working with larger SoCs. Plus it’s cheap and you can see the connector pins without a microscope.
>Somebody get this man an Arduino Experimentation Kit, stat! ;)
I hear you, but I’m way more attracted to the BeagleBone. Running Linux is something I understand, and I’m confident that knowing the toolchain would give me a big leg up. The Arduino and Raspberry Pi fans are right to fear it.
I don’t think many Raspberry Pi fans even know the BeagleBone exists, much less fear it. Cost and network effects (community support and complementary products) are currently strong BeagleBone disadvantages. Even though on paper the Beagle is mostly a better machine, the Pi is good enough for a lot of use cases, and the community support seems much stronger for Pi.
Whatever happens, I expect the battle for supremacy to be long and interesting.
I use all three boards, but my Pis outnumber my Beagles 3:1, and I have a drawer full of AVR chips. When the Pi came out, and they said I could get $160 of Linux computer for $40, I said, “Great! I’ll pay $160 and put Linux in 4 different places at a time.”
The Pi runs Debian supported by the hardware vendor, so every Pi is just another Linux server. Beagle picked Angstrom, which has a nasty learning curve, as well as egregious performance requirements for any machine that aspires to be a cross-building host (e.g. the packaging tool wants 8GB+ of RAM to build hello-world). There seem to be more Pi accessories ready to buy than Beagle ones, and IMHO they tend to be more practical designs, too.
>Cost and network effects (community support and complementary products) are currently strong BeagleBone disadvantages.
Whereas, while I totally loved the idea of the Raspberry Pi, absence of an onboard clock is a flat showstopper for almost anything I might personally want to do with it. That is, assuming I develop the hardware skills to play. Which I have no doubt I could do if some particular project motivated me enough.
Also, the BeagleBone is purer from an open-source-all-the-way-down perspective. Fewer potential IP landmines in the design makes me feel more comfortable about downstream risks,
One of the interesting things to watch about both machines is how fast a supporting ecology developed around them. I don’t know if this will or can be true for a third competitor; it feels to me like BeagleBone has crossed a line to “good enough” and future improvements in this product area will be less revolutionary, more incremental.
>Beagle picked Angstrom
The BBB runs Ubuntu now. Can’t get more mainstream than that.
@esr:
> My “ill-founded idea” turned into an inexpensive and excellent time source, delivering 1ms accuracy (not good enough for Stratum 1 but 10 times better than typical Stratum 2) for $33 a pop. Nothing even remotely as good is available for a price less than an order of magnitude higher. We could use a lot more such ill-foundedness, I think.
This “ill-founded” idea is a fundamental contribution for mankind: It eliminates “distributed-computing” and all its shit algorithms for good and for eternity.
@Zygo: I’ve just learned that I got something wrong earlier. There still is direct hardware support in ntpd, not that you get any clue about this from the manual page. I’m trying to find out more.
Funny you should mention clocks…
I built a clock (i.e. a wall-mounted appliance that informs you of the current time by controlling a lot of LEDs) using a Raspberry Pi and leftover parts from other projects. It gets time from NTP over Wifi instead of using a local RTC, because a Wifi module is $3 cheaper than a bad RTC module, and half the price of a good one. The clock display changes color depending on various NTP state parameters.
Services like OSH Park and widely available open design files mean that simple expansion boards can be designed and built with a laptop, a credit card, and an Internet connection. We’re well past the days of having to have your own chemical lab to make boards, and the recent SBC designs are laid out to make custom boards easy and useful.
I expect a de facto standard form factor for expansion connectors to emerge, making expansion boards become portable between SBC host boards. There are Raspberry-to-Arduino shim boards already, and I might build my own BeagleBone-to-RaspberryPi shim for project I’m working on (I’ve built a non-trivial expansion board for the Pi, but I’m rapidly outgrowing the Pi’s CPU, and I’m not rebuilding the expansion board just because the BeagleBone connector happens to be a different physical shape).
I added some information about the new no-soldering GPS/PPS module for the Raspberry Pi here:
http://www.satsignal.eu/ntp/Raspberry-Pi-NTP.html#no-soldering
I also made a digital wall clock, although different from Zygo’s design:
http://www.satsignal.eu/raspberry-pi/DigitalClock.html
Cheers!
ESR said: Whereas, while I totally loved the idea of the Raspberry Pi, absence of an onboard clock is a flat showstopper for almost anything I might personally want to do with it. That is, assuming I develop the hardware skills to play. Which I have no doubt I could do if some particular project motivated me enough.
I got a Pi just as a random toy, and decided I wanted an RTC. So I got one from Britain for like $15.
Of course, might not be good enough for your project-space at the moment, but just like my PCs run ntp clients to keep time, so does the Pi, when it’s on. The RTC is “just because” for me.
(If I ever get serious about Embedded Stuff, I’m sure I’ll get a NETduino; I’d rather buy “too much” controller and “waste” CPU, to leverage my .NET skills and avoid ever having to write another line of C, or deal with Python.
But that’s just me.)
Anyone have an original 256 MB RPi Model B they want to unload? I’ve got a cabinet with three of them, and blew one up by hooking the power to it backwards. (Grumble.)
I haven’t looked at the BBB at all. If it’s a good RPi competitor, that’s a good thing all the way around.
In any case, deploying tiny SBCs as single-purpose servers is just the next step along the path that led from boxes to blades…
ESR: would you care to comment on the progressive closed-sourcing of Android?
>ESR: would you care to comment on the progressive closed-sourcing of Android?
Yes. If it ever becomes a real problem, CyanogenMod will fix it.
> […] blew one up by hooking the power to it backwards.
It isn’t protected against original Murphy law: “If there are two or more ways to do something, and one of those ways can result in a catastrophe, then someone will do it.”?
BTW., slightly off-topic – what do you think about Microduino Kickstarter^W project http://www.microduino.cc ?
Jakub: “It isn’t protected against original Murphy law: “If there are two or more ways to do something, and one of those ways can result in a catastrophe, then someone will do it.”?”
Nope, not if you’re hard-wiring power to it through the +5/ground pins on the GPIO connector instead of the usual micro-USB power connector. The lack of a separate +5 input that’s not the micro-USB is one of the RPI’s shortcomings.
Our host will be even more impossible now. He’s got his code in a Benz!
http://www4.mercedes-benz.com/manual-cars/ba/foss/content/en/assets/FOSS_licences.pdf
>Our host will be even more impossible now. He’s got his code in a Benz!
Meh. I’m a geek. I still think GPSD’s coolest deployment story is robot submarines.
> If it ever becomes a real problem, CyanogenMod will fix it.
Assuming it’s not already a problem (opinions differ), CM may, or may not be able to fix it.
The motorola boot loader still isn’t unlocked.
Samsung doesn’t seem interested in supporting CM.
The CM ‘newco’ is venture-backed, and that is typically death for any open source effort, as the VCs eventually decide to turn the company toward profitability.
> I hear you, but I’m way more attracted to the BeagleBone. Running Linux is something I understand, and I’m confident that knowing the toolchain would give me a big leg up. The Arduino and Raspberry Pi fans are right to fear it.
RPi probably. Arduino I’m not as sure about, mainly because they feel aimed at different components of the same market. There are some control tasks I see arduinos used for which would be similar to the old nuclear warhead on a mosquito analogy if you put a BBB (or RPi for that matter) to the purpose. The power usage of the BBB (Their website suggests a 10W supply for the 5V rail) is also much higher than some arduinos. They would have more to fear from a decent low power ARM cortex M0+ board.
I used to work at a robot submarine company. To be honest I was surprised we didn’t use gpsd.
@esr
> Yes. If it ever becomes a real problem, CyanogenMod will fix it.
That is easy to say, except for the fact that phones, by their nature, have to connect to infrastructure. And people who run infrastructure don’t do it open source, and frequently say things like “it must pass our certification test before you can connect it to our network” via some sort of signing process. And it is often easy to justify it in the name of network stability, preventing piracy and preventing “hacking”.
So I think your confidence is a little misplaced.
My view is that the whole software world is fast turning into a walled garden. One of the big things in corporate America today is locking down PCs really tight so that you can’t install software, and delivering the required Apps through a corporate App store. This is the legacy of Steve Jobs, for good or ill.
>My view is that the whole software world is fast turning into a walled garden. One of the big things in corporate America today is locking down PCs really tight so that you can’t install software, and delivering the required Apps through a corporate App store.
So conditions inside the prison are getting worse. Fortunately, the prison has an outside, which is getting more various and harder to control all the time. All the activity around the Arduino, Raspberry Pi and Beaglebone is an indicator of this.
What do the little SBCs have to do with corporate software, which is what Fluffy Girl was talking about?
>What do the little SBCs have to do with corporate software, which is what Fluffy Girl was talking about?
My point was that the corporate prison is not “the whole software world”. She mistakes local overcontrol for global stagnation because her sample is limited.
esr on 2013-10-25 at 13:42:53 said:
> My point was that the corporate prison is not “the whole software world”. She mistakes local overcontrol for global stagnation because her sample is limited.
No doubt my world view is parochial. However, what I observe is that tablets and phones are becoming one of the central computer resources touching people’s lives and they are very prone to this walled garden effect that I was talking about. In particular, even with an open Android, it still has to connect to a closed phone network, and that closed network can apply restrictions on that open Android. “Only approved versions of Android can run our on network.” Especially in the USA where most people still buy their phone from their carrier.
And do I really need to point out the constant push to get proprietary and patented material into Web standards?
>In particular, even with an open Android, it still has to connect to a closed phone network, and that closed network can apply restrictions on that open Android. “Only approved versions of Android can run our on network.”
I hear talk of this but it hasn’t affected me once in four Android phones. I buy bring-your-own Google devices like the N4 – no bloatware, and how can the carrier restrict me? They didn’t build the image on my phone!
Sure, if you buy a heavily subsidized crap-phone the carrier can try to pull tricks. They seem to be remarkably ineffective at it, however. The attempts to block tethering have failed quite miserably. About the only power they seem to have is the ability to install bloatware you can’t remove. They can’t make you actually use it.
>And do I really need to point out the constant push to get proprietary and patented material into Web standards?
Which has failed. The best they’ve managed is that wacky extension thing for proprietary video codecs in. Which will accomplish approximately fuck-all because YouTube.
They approved the radio firmware, which you probably didn’t build, and which carriers can selectively turn off. The radio firmware in the phone and in the carrier base stations have a signature or certificate for every carrier/manufacturer they work with. Minimally it’s a list of IMEI patterns to match, but it can also include various proprietary security measures up to and including mutually authenticated PKI certificates. Admittedly, it boils down to the remote attestation problem, so you can get around carrier-side filters using a certificate ripped from any working phone–but that’s not a solution if you want to build new phones instead of cannibalizing old ones.
Phones tend to have two processors: one for radio signalling and one for application software. The radio processor often supervises the application processor that Android runs on. Among other things, the radio processor has control of power, so it can turn itself on whenever it likes. It also can have access to the same RAM as the Android processor (sometimes more access if there is hardware DRM on the platform). One RAM module is cheaper than two, and the only security threat the mobile phone industry spends money on is its own subscribers.
The radio firmware tends to have enough holes that you should assume everyone competent within 6 miles of you who wants root on your phone probably already has it by now, and since they can do it without being visible from Android, it’s hard to tell when it happens.
Some phones separate the two processors with a high speed UART or SPI bridge. It reduces the attack surface from “read/write access to all the RAM” to something smaller, but it doesn’t guarantee there isn’t some stupid vulnerability in there somewhere (e.g. shared firmware ROM with write access from both processors). It also dramatically increases the cost of the phone, since the hardware now looks like a Beagle with the guts of a USB cell modem stuck into it, and it has battery demands to match.
To be fair, your carrier and phone manufacturer have lots of reasons to not remotely mess with the software on your phone in any end-user-visible way–but only those entities have those reasons. Everyone else has no particular disincentive, as long as they can get away with it.
There’s also no reason to assume that some manufacturer won’t take a page from the video game console playbook, and decide that future radio firmware updates, over which users may have no control, will summarily wipe customized Android images and any other data they don’t like off of “their” phones some day.
Sure they can, they just turn it on remotely, or unconditionally.
You mean the video hosting service patrolled by revenue-seeking copyright robots? There’s a fair amount of DRM already in YouTube buried in the Flash player. You tend to run into it rather hard when you try to collect music videos.
>Sure they can, they just turn it on remotely, or unconditionally.
You seem to be talking about a different problem than I or Fluffy Girl is. By your account they could possibly put filtering or monitoring software in my phone, but they can’t make me (say) use Facebook. And with all the leverage you think they have the carriers still couldn’t block tethering, so there is something in the techno-legal surround that prevents them from exerting it effectively. Probably it’s that the radio firmware is built by people the carriers don’t own, so that even though they approve it they’re not able to Trojan-horse it.
>The radio firmware tends to have enough holes that you should assume everyone competent within 6 miles of you who wants root on your phone probably already has it by now,
Which is why I limit my exposure by not putting any privacy-critical data on my phone. But this is not an argument that the forces of reaction are inevitably going to crush open source, nor is it even a death knell for CyanogenMod. And CyanogenMod has some important top cover; it’s a key though unacknowledged piece in Google’s plan to commoditize the carriers, so Google is not going to let it die quietly.
@esr
> hear talk of this but it hasn’t affected me once in four Android phones.
I’m not saying it has, I am saying that your confidence in open source to save you from that in a system that includes non open components is excessive. Better Android than Apple, but still there is both a serious risk of lockdown and plenty of reason to imagine it will happen. The average public is much less concerned with this than people who care about W3C standardization processes (as is evidenced by the shoulder shrug of interest in the outrageous spying that Snowden has revealed.)
> and how can the carrier restrict me?
Pretty easily… Every five seconds during a call demand that the phone send back a selected series of 256 bytes from the image signed with a hash. If CyanogenMod adds that feature, the DCMA puts them out of business.
All the tools are there, and the necessary social pressure to prevent it from happening is not.
I remember reading years ago a piece by one of my favorite writers, Bill Freeza called the crucible of capitalism in which he argued that the new freedoms of the internet would emasculate the power of government to drown us with their power and would lead to a new blossoming capitalist, liberal society. I drunk the kool aid. I am disappointed though. It seems these selfsame tools have been turned on us and robbed us of our privacy, and the integrity of our property or ability to transact freely.
We live in a society today that is so used to the idea that our privacy is gone that we read this article without batting an eyelash.
http://dailycaller.com/2013/10/25/exclusive-feds-confiscate-investigative-reporters-confidential-files-during-raid/
(Summary, the cops executed a warrant to confiscate guns and ammo from a reporter’s home. They instead confiscated all her private notes and files from her whistleblower sources. She has been a constant critic of the TSA and their practices, and it just so happens that the cops decided to turn over those notes, including the sources, to the TSA to “check that she had a legitimate right to have them.” The pretext being that it included a few official papers obtained under the FOIA.)
No doubt the wistleblowers will be looking for a new job. The cops who utterly violated this woman’s rights, and utterly undermined he process whereby the public can keep the government in check, will undoubtedly receive praise and commendation.
>Pretty easily… Every five seconds during a call demand that the phone send back a selected series of 256 bytes from the image signed with a hash. If CyanogenMod adds that feature, the DCMA puts them out of business
The FCC forbids this sort of jiggery-pokery under the Part C restrictions that Google got the carriers stuck with in the last major spectrum auction.
>We live in a society today that is so used to the idea that our privacy is gone that we read this article without batting an eyelash
The Washington Times is suing over this, quite properly alleging civil rights violations. And there’s a massive, bipartisan anti-survellance rally going down in DC today.
I’m not saying everything is rosy, but our case is not quite as desperate as you suggest. Remember that while the carriers and other anti-open-source groups such as the MPAA are malevolent and powerful, they are also deeply stupid. As I pointed out earlier, the carriers couldn’t even block tethering!
If T-Mobile wanted to block tethering, they could. Instead what they do is detect it and automatically, silently add a huge-ass surcharge to your cellphone bill if you do it..Why block tethering and sell it as an add-on when they can charge the same amount without you knowing or thinking about it?
>If T-Mobile wanted to block tethering, they could. Instead what they do is detect it and automatically, silently add a huge-ass surcharge to your cellphone bill if you do it
[citation needed]
I’m a T-Mobile customer. This has not happened to me.
Video games were thought to be a passing fad by 1984. Then in 1985, along came Nintendo and turned them into a walled garden. Now they are a multibillion dollar industry where each major game gets a budget to rival Hollywood blockbusters.
To the vast majority of people out there, walled gardens are a feature. They provide a consistent experience for all, filter out malware by the only strategy known to work — whitelisting — and drastically reduce configuration headaches and support calls.
And PCs were locked down well before the iPhone. There were Dilbert cartoons about it. One of the best implementations even came from Apple: the “At Ease” shell for the Mac which supported multiple users, each with their own set of approved applications and private file space. If anything, the advent of the iPhone is what started encouraging corporate IT to implement BYOD policies — high-level execs wanted to use their shiny new iPhone or iPad instead of the company-issued BlackBerry or ThinkPad.
Yes, and left-wing rags like Mondo 2000 promoted the same peace-and-harmony message from a socialist perspective. Even R.U. Sirius admits what a mistake he made because — big surprise — when you let TPTB build out the infrastructure, TPTB own and control the network.
See for example:
http://forums.androidcentral.com/google-nexus-4/264791-tethering-tmos-new-plans.html
It seems that if you have an IMEI T-Mo doesn’t recognize, you might be able to get away with some under-the-radar tethering without paying for it — until your data usage spikes and they start doing deep-packet inspection. Since all this is of course T-Mobile trade secret, we have no way of knowing for sure what their actual policies are. But the technology to detect whether packets originate from the device itself or from a PC using the device as a proxy exists, and carriers have an interest in deploying it.
The situation may have changed again, as it seems Tmo is now offering unlimited plans with 2.5GiB of free tethering included. But that’s market forces at work, not a lack of capability to detect and block or bill for tethering.
@Jeff Read
The EU regulations on carrier licenses forbid most of your “carrier policies”. Attempts in the Netherlands to use deep packet inspection to charge for Whatsup and Skype usage were struck down immediately by the regulators.
Any attempt to block Cyanogen would be illegal as far as I know. Just as unlocking phones is legal and the carriers have no say in what phones are used on their networks.
I have not heard of any attempts to block tethering by EU carriers. But I might have missed it.
I forgot to add a “@Fluffy Girl” to my previous comment.
You don’t have to tell me that Europe has solved with government regulation problems that have proven intractable for the sacred free market. I find it interesting how Eric keeps talking up Google’s Cunning Plan to commoditize the carriers, when the same has been done for years in Europe with the one tool that’s complete anathema to Eric: government regulation.
States are beautiful things, Eric. Learn to love them.
Anyway, I don’t know about blocking tethering, but as late as 2011 at least one UK carrier (O2) has tried to tack surcharges on for it using deep-packet inspection. That has since been rescinded, although I don’t know if it was due to regulatory pressure or market forces (the latter of which also work a lot better in Europe in the telecom industry, thanks to government regulation).
@Jeff Read
>You don’t have to tell me that Europe has solved with government regulation problems that have proven intractable for the sacred free market.
See here is the problem we hear again and again about so called “failures” of the free market. The wireless market is far from free, especially in the context of bit hauling. The basic medium that is needed for bit hauling is not owned by the companies using it, it is “owned” by the government, who pimp it out for various purposes, for large rents, and with beacoup attached conditions. Add on top of that the regulatory body of the FCC, with many very silly policies, and you end up with a half assed blend between regulation and non regulation.
So it is crazy to say that the US wireless market is unregulated. It is, heavily. WiFi on the other hand is not, and it is spectacular success with costs spiraling down almost as fast as functionality spirals up. (I have argued here before that WiFi is essentially private property in the spectrum — see my many previous comments on this matter.)
Claiming the wireless market is a free market is pretty much as laughable as the present argument that the “unregulated, free market” US healthcare market is a disaster, that needs to be replaced by the dishonestly named “Affordable Care Act”. As if LACK of regulation is the problem with medical care in the USA. (I have worked in the production of medical devices — Forests are petitioning the government for relief from the burden it places on them.)
BTW, why has our press not pointed out the fundamental principle of Obamacare — namely the principle that the poorest members of society, the young, should subsidize the richest members of society, the old? I suppose since we already do that with Social Security and Medicare, it isn’t much of a jump.
BTW2, since this is an educated audience that understands these things, how exactly is it possible for the government to spend $640million dollars on that web site? I mean I understand that governments are profligate — but it takes a level of profligacy that would put the Bourbon monarchy to shame to spend that sort of money. We here all know that that is at least 100x as much as it should have cost. And it doesn’t work. However, fear not, they plan is to throw more bodies at the problem. Fred Brooks is pulling out whatever hair he has remaining.
And here is the latest news — they do not have a cyber security policy in place. Because after all, that web site is VERY unlikely to be the subject of attack… right?
These are the people who want to decide if you can get a heart transplant… God help us all.
Oh, no — WiFi is regulated: there are restrictions on frequency and transmission power. What it is not is licensed, and that’s because there are restrictions on transmission power, and the signal deteriorates to nothing at smaller-than-city-block scales.
Anyone who thinks an unregulated spectrum is viable really needs to learn how radio waves work. Spectrum allocation is really just another form of right-of-way or easement. This really has nothing to do with the kind of regulation I’m talking about.
In Europe, there is one cellphone standard — GSM. All carriers must provide a GSM network on certain standard frequencies. In addition, European carriers are forbidden from employing some of the predatory business policies that USA carriers use: carrier-locking phones, onerous contracts, predatory pricing including charging different data rates based on content or charging for SMS at all, etc. They are forced to compete on equal terms and serve as generic bit-haulers who may be connected to by any compliant GSM phone, and that is exactly the ideal situation.
Oh, I agree. Obamacare is fucked; I should know, I’ve been living under its trial balloon, Romneycare.
The obvious solution to the problem of Obamacare is the one that Europe — and literally every other developed nation except the United States — has adopted: namely, the government should pay everyone’s medical bills, no questions asked. It turns out that if you replace the maze of provisos and quid pro quos with a simple handout, you save more money in overhead costs than you would lose in fraudulent claims. That’s part of why the guaranteed basic income is becoming a popular policy.
> the government should pay everyone’s medical bills, no questions asked
which eliminates competition, and prices rise.
@R. Duke
I seem to remember that the USA spends more on health care than anyine else in the world. Both in absolute terms per capita as well as percentage of GDP.
The NHS in the UK has the lowest costs of them all.
Europe rations care. Those who can, pay to go around the rationing. Many of those who canot or do not pay to go around the rationing do not realize what is happening.
My father-in-law got colon cancer. He had to wait months for a diagnosis, months for surgery, and months afterward before they started chemotherapy. The result should not suprise you.
Contrast that with my father, who got his surgery within weeks of the diagnosis and did not require chemo. His colon cancer never got a chance to kill him.
The plural of anecdote is not data, but a well off Swedish Finn who trusted the system did not get the care he needed.
@Jeff Read
> Oh, no — WiFi is regulated:
In the same way private property is. Certainly physical limits are placed on WiFi. Of course the nature of radio means that fences and lines on maps don’t work. So we do the equivalent, limiting frequency and power, the former representing where, the later how big. Of course the lines aren’t as bright as a boundary on a map, But that is the nature of the beast. Within my apartment I have free reign as to what data I send in that frequency range, out to a certain distance. Since power roughly corresponds to the physical volume I get to control that spectrum within. I get the choose what protocol to use, I get to decide how to distribute the packets between video and data. I get to choose what level of net neutrality to use.
Which isn’t to say there isn’t any interference. There is. I live in an apartment complex and there are dozens of wifi signals going on at various power levels. However, somehow I don’t need the FCC to come in and say that I get channel 1, while my neighbor gets channel 2, and the guy down the hall channel 3.
It is as close to spectrum ownership as you can get with radio, and it has proved spectacularly useful.
@Jeff Read
>The obvious solution to the problem of Obamacare .. namely, the government should pay everyone’s medical bills, no questions asked.
No questions except questions like “are you too fat to have a hip replacement.” I find the idea of begging the government for access to one of life’s most precious necessities, utterly horrifying. I’m not a fan of the way the medical insurance industry is run currently either, but I’d rather have an enforceable contract rather than a government favor. The problems in the medical industry are mostly due to the current interference of the government, not too little interference. The problem is the same pseudo free market where it is hard to tell em to “go to hell” if they don’t perform.
I’d remind you, they can’t even get their web site for that impossibly complex task of “filling in forms” right.
> It turns out that if you replace the maze of provisos and quid pro quos
Do you seriously contend that Obamacare is free of a maze of provisos and quid pro quos? Obamacare could almost be defined as “a maze of provisos and quid pro quos.”
> with a simple handout,
Medical care is a limited resource. Zero cost does not work well for limited resources. It always requires some “wise” person to decide how the resources are allocated. They almost always do it badly. They almost always do it corruptly.
>No questions except questions like “are you too fat to have a hip replacement.”
There are only two ways health-care resources can be allocated. By cost, which means the poor suffer. Or by bureaucrats executing regulations, in which case the poor still suffer and so does anyone that the bureacrats find undeserving.
@Winter
You may be correct in your point , but comparing per capita or percentage of GDP is not the correct way to support it. As people and societies grow wealthier, there are more funds available to spend on healthcare. The increase in per capita spending or as a percentage of GDP could be due to a choice to spend more funds on healthcare due to a wealth differential or to different perceived marginal utilities.
@Jeff:
>You don’t have to tell me that Europe has solved with government regulation problems that have proven intractable for the
sacred free market.
Very often the problems that are solved by a given regulation are caused by some other regualtion. For example, copyright and patent laws grant certain monopolies to the holders of copyrights and patents. While the law remains as it is, the US government has the duty to make sure that the monopolies it has granted are not abused, which in practice means being more proactive about taking anti-trust action against the likes of Microsoft and Apple (among others). This could take the form of stronger antitrust law, but actual enforcement of existing antitrust law would be enough in my view (US v. Microsoft was a farce in that regard). But a good argument can be made that it would be better to get rid of copyrights, patents, the DMCA, etc, and not take antitrust action against anybody.
>Anyone who thinks an unregulated spectrum is viable really needs to learn how radio waves work.
One big question is whether whatever regulation exists is enforced with prior restraint. If the firmware for a computing device controlling a transmitter has to be approved by state regulators ahead of time (as I understand to be generally the case for cell phones), that creates barriers to entry that work in favor of large, entrenched players in the market, which then brings us back to the above paragraph (yes, if we’re going to require pre-approved firmware, certain regulations are required to prevent abuses of the oligopoly thus created, but it might very well be better not to require pre-approved firmware).
Frankly, I don’t see why radio waves need much more regulation than sound waves. If music from a wild party disturbs the neighbors, the police get called and people get written up for disturbing the peace, but AFAIK there are no legal restrictions on how loud a stereo can be. Why should it be any different for radio? No transmission power restrictions, but if your wild LAN party kills WiFi reception for the whole block, the neighbors call the police and you get fined for disturbing the peace.
@Fluffy Girl RE: Obamacare and single payer systems:
Jeff is saying that a single payer system would be preferable to Obamacare, and I agree with that. At the same time, deregulation would be preferable to either.
@esr
> Or by bureaucrats executing regulations,
There is a third possibility (arguably a subtle variant on the above), namely a charitable structure where the poor who can’t afford a necessity are provided it by the generosity of others (such as charity hospitals, or doctors waiving or reducing their fees for the benefit of their patients.) This works very well usually because the providers of charity are close enough to the recipients that the usual abuses of other people’s generosity is much less common.
it is a world of difference between the kindness of strangers giving a leg up to the poor an unfortunate, and an entitlement program run by people who have no skin in the game.
“To proud to accept help” or the “stigma of charity” are extremely effective mechanisms to get people to provide for themselves, but have a safety net if they just can’t. I hate entitlement programs, but I am a total sucker for real hard luck stories.
Of course it is hard to do a lot of that because government regulation, and miscellaneous other government actions, makes medical care so insanely expensive.
But there’s no way of effectively regulating radio transmitters without an approval process for the firmware since the firmware, among other things, controls the output power of the radio. This is part of why open-source wifi drivers have been so hard to come by; the FCC considers the radio and any software controlling it to be a unit when it approves the device for unlicensed use; if you put different firmware on it, the radio falls out of spec and you can’t sell it, because if you did, you are providing a device that might possibly transmit above unlicensed power maximums, and therefore would require a license to operate. So what Atheros did was to release binary-blob Linux drivers and keep the source to the relevant interfaces closed, thus ensuring that what they sold was only nominally capable of acceptable power outputs. Now the OpenBSD folks came along and said “fuck you, we’re Canadian” and released a fully OSS driver for Atheros gear, but that’s a different issue.
And you can’t have Jessica Boxer’s property lines for radio waves without these regulations.
Manufacturers can get around this by designing the radio hardware to never exceed a given power maximum and only operate within certain frequencies, thus preventing the need to regulate the firmware since the firmware can’t command any out-of-spec power or frequency. But this approach greatly increases the cost of the hardware. So regulating the firmware actually provides a means for manufacturers to stamp out cheap devices which would otherwise be not approved to operate unlicensed at all, makes the development of cheap smartphones possible, and lowers the barrier of entry for new players in the market.
@Jeff:
>But there’s no way of effectively regulating radio transmitters without an approval process for the firmware since the firmware, among other things, controls the output power of the radio.
But as I see it the goal is not the regulation of transmitters themselves, but rather the regulation of the actual act of transmission.
>This is part of why open-source wifi drivers have been so hard to come by; the FCC considers the radio and any software controlling it to be a unit when it approves the device for unlicensed use;
And that is exactly what brings the current regulatory regime into question. If you can’t have open source drivers or firmware for hardware in a given industry due to regulatory restrictions, then you don’t have a free market, so when the resulting oligopoly or monopoly causes problems and you solve them with more regulation, you can’t say “we solved problems with regulation that were intractable to the free market”.
> if you put different firmware on it, the radio falls out of spec and you can’t sell it, because if you did, you are providing a device that might possibly transmit above unlicensed power maximums, and therefore would require a license to operate.
Once again, I’m more or less fine with requiring a license to operate above a certain power level, but not for manufacturing/importing/owning a device capable of exceeding that power level.
>And you can’t have Jessica Boxer’s property lines for radio waves without these regulations.
Yes you can. You just prosecute people for trespassing, rather than confining them to their own property with lock and key so they *can’t* tresspass.
Once again, why do radio waves need more regulation than sound waves? Human speech is limited to only one channel, and for various reasons noise when one is not trying to talk or listen is more of a problem for humans than for radio devices, but I am not routinely bothered by loud drunken arguments despite the fact that there is comparatively little regulation of sound noise (not even to the point of Jessica’s “property lines”) and that people are not required to have their vocal cords surgically altered to render them incapable of shouting. People still generally manage to keep the volume of the sound they create down to acceptable levels, and those who don’t get fined for disturbing the peace *after the fact*.
@BobW
“The plural of anecdote is not data, but a well off Swedish Finn who trusted the system did not get the care he needed.”
I have both data and anecdotes. Here is the data:
http://www.commonwealthfund.org/~/media/Files/Publications/Issue%20Brief/2011/Jul/1532_Squires_US_hlt_sys_comparison_12_nations_intl_brief_v2.pdf
Conclusion: The USA spends most and wastes even more for a mediocre health care system.
My anecdotes. I know far too many women who got breast cancer. They were all helped with the full gamut of surgery, chemo & radiotherapy, plus years of adjuvant therapy. All on basic medical state insurance. No extra money needed. Even better, I know people working at one of the best cancer research centers in the world, and they work the full care on everyone. Neither rationing nor any quality difference based on money.
“Breaking bad” could only be conceived in the USA.
http://www.buzzfeed.com/lukelewis/if-breaking-bad-had-been-set-in-the-uk
@Fluffy Girl
“No questions except questions like “are you too fat to have a hip replacement.””
I cannot comment on other countries, but over here in the Netherlands I know a lot of people who got hip and knee replacements, no questions asked, no extra payments needed. And most were old people. However, if you are too heavy, a hip replacement might not “stick”, which is a medical condition.
@esr
“Or by bureaucrats executing regulations, in which case the poor still suffer and so does anyone that the bureacrats find undeserving.”
Which is a bald-faced lie. In those countries where there is rationing, it is never about “undeserving people”. Again, in the Netherlands (where I have experience) the only rationing ever considered is quality-adjusted life year (QALY) versus costs. Which is a purely medical qualification (look it up). And even these are never used against old people (any suggestion of such use leads to a political shit storm).
@Joe Presley
“As people and societies grow wealthier, there are more funds available to spend on healthcare.”
That is called “waste”.
@Fluffy Girl
“BTW2, since this is an educated audience that understands these things, how exactly is it possible for the government to spend $640million dollars on that web site?”
The seven deadly sins of HealthCare.gov
http://arstechnica.com/information-technology/2013/10/the-seven-deadly-sins-of-healthcare-gov/
“Once again, why do radio waves need more regulation than sound waves? (snip reasons)”
Sound does not carry over nearly the distance that radio waves do. It’s also impossible to generate (by voice) an excessively loud sound that does not have any effect on the person generating it, because we cannot turn off our ears.
Random832
Try generating over 120 dB in your city block. See how long it takes before they shut you down.
And, actually, we do turn off our ears while speaking. More precisely, we dampen the transmission (by up to 30dB).
https://en.wikipedia.org/wiki/Stapedius
@Winter
> In other words, HealthCare.gov’s rollout was almost identical to every Web-based ERP or CRM rollout at a major corporation.
That is plainly ridiculous. I have been involved and observed many high volume web sites that worked well, if initially a little glitchy. I have used the web sites of various medical insurance providers and they work just fine.
It is easy to excuse a royal cock up by saying, “everyone is screwed up”, however, the truth is everyone isn’t screwed up. Far more complex web sites work every day without any problems. Sure there are a lot of users, but none of them are doing much more than filling in forms and reading CMS content. The hell with their stupid web site, they should have used Joomla.
My really questions are this:
1. How could you screw it up so badly that it isn’t just glitchy, it really doesn’t work at all?
2. How could they POSSIBLY spend that much money on a web site with a fairly simple purpose? Remember, I think it cost at least 100 times what it should, you know 10,000%
3. How can they possibly think that their solution, in face of Brooks law, is going to fix it?
4. How can they POSSIBLY countenance delivering this web site without an AGRESSSIVE cyber security and PII security policy and framework?
On this last point alone, someone should be going to jail.
@Jeff Read
>But there’s no way of effectively regulating radio transmitters without an approval process
Drunks pee in alleyways. How can we prevent that? Perhaps we should set up the FUC (Federal Urethra Commission) to regulate our peeing activity? It could regulate the amount of liquids provided to people at restaurants, and perhaps work with local water authorities to manage how much drinking water is supplied to households. We could require households to have special plumbing for potable water, and put some sort of bad taste chemical in other water for showers, toilets, hand washing etc. Surely that would go a long way to preventing the “peeing in alleyway nuisance.”
Or alternatively we could just bust the drunks when they whip out their dicks… (As Queen Victoria allegedly said of lesbianism “Women do not do such things…”.)
@Fluffy Girl
“My really questions are this:”
Some of your questions are answered by the linked article. I just posted it to help you answer you questions. I have no idea how and why they failed. I even have no idea what they were supposed to do in the first place.
Winter,
CRM and ERP deployments are usually low-volume, internal to a company network. They’re not the standard that HCgov has to meet. Google, Facebook, and Twitter are.
Like most enterprise software, HCgov is a shit sandwich because the buyers are not the users.
I find it amusing how Obama manages to assemble a crack team of Silicon Valley’s finest to run the IT for his campaign but when it comes to his flagship program, the best he could do is throw more money at B and C players.
That’s the Obama presidency in a nutshell. He’s ruthlessly efficient at acquiring power but once he’s got the power, he falters on executing his promises.
@Jeff Read
> That’s the Obama presidency in a nutshell. He’s ruthlessly efficient at acquiring power but once he’s got the power, he falters on executing his promises.
That is a great analysis.
@Jeff Read
Maybe the website was just an aftertought?
@Winter
As too many others have said, the USA has the worst of both worlds. It does not have free market healthcare, and it does not have single payer.
A great many of us doubt the claims that single-payer works well for everybody. Somebody must be falling through the cracks. In the USA those cracks will be enormous. Organizations do not scale well. The USA would be smart to leave such matters to the states.
@Jeff Read
Napoleon’s Dictum: “Never ascribe to malice what can adequately be explained by incompetence.”
These days I wonder. Once is chance, twice is coincidence, three times is enemy action.
It’s part of the health insurance marketplace provisions of the ACA and an essential part of the law. Remember, I’ve lived through, and availed myself of the services of, Obamacare’s direct predecessor in Massachusetts — which was signed into law by Obama’s later political rival, Mitt Romney, in 2006 and went into effect in 2007. The effects are broadly similar: a requirement that everyone be covered, a restriction of insurers on whom they must not refuse coverage to and what they must cover, the setup of an online Web site to help uninsured people find affordable care packages, and government aid for insurance for the poorest individuals. I know, to a degree that many Americans let alone foreigners don’t, how shitty Obamacare really is in practice. Though the Massachusetts state online insurance marketplace worked fine for my needs, the “affordable care” I purchased sucked a donkey’s ass and cost close to $500 a month. Though I think the hearts of the architects of Romneycare and Obamacare may be in the right place, this measure really isn’t enough. Health insurance companies make their money by getting you to pay without actually covering you. Now that they have to cover certain people who cannot afford their payments, they’re keeping their balance sheets up by punishing everyone else.
@Jeff Read
>I know, to a degree that many Americans let alone foreigners don’t, how shitty Obamacare really is in practice.
But this is the most shocking failure of all. The whole point of the federal system is to be able to try different ways of doing things in different states, as appropriate for different populations. Which is why the global sucking sound as EVERYTHING gets sucked up to Washington is such a dreadful thing. To implement Obamacare without knowing how it will go is silly, to implement it knowing where it will go (by analogy with MA) is positively mendacious.
> Health insurance companies make their money by getting you to pay without actually covering you.
This is like saying I make money by creating software systems with bugs, so that they need me around to fix them. The truth is that insurance companies provide a risk pool, and take a profit to compensate them for their part of the risk, and for the cost of managing the pool. If the government forces them to accept riskier people into the pool then everyone’s risk is increased and everyone has to pay more. The only people doing any punishing are the government who are interfering with their risk models.
Perhaps we as a society want to find it taboo to charge different rates for diabetics than for non diabetics, or women with particular breast cancer genes than not, and make it illegal for the insurance companies to factor that into their risk tables. But if we as a society decide to do this, we have no right to call the Insurance companies evil should this mean an average increase in premium for everyone. Math is immune to the manipulation of emotion.
I read an interesting article BTW by Robert Cringely. He contends that much of the problems with modern healthcare (and specifically the denial of coverage to higher risk patients) is due to the facility of big data analysis that has appeared in the last few decades. Which is to say we can exclude them because we have to computing power to reduce the size of our quantum of risk analysis. I think it is an overly simplistic analysis. Most of the problems with insurance are created by government regulation. However, it is an interesting addition to the debate.
Should the government mandate ignorance as part of the insurance company’s risk models?
http://www.cringely.com/2013/10/26/big-data-destroying-u-s-healthcare-system/
Of course the truth is that the whole thing is a big bait and switch to destroy private medical insurance and force everyone into a single payer federal system. So much for too big to fail, this is specifically too big so that it will fail.
@esr
> Meh. I’m a geek. I still think GPSD’s coolest deployment story is robot submarines.
Please tell me they are going to use GPSD in this:
http://news.slashdot.org/story/13/10/30/025224/drone-mounted-laser-weapons-are-on-the-way
>Please tell me they are going to use GPSD in [DARPA’s laser-equipped drones]:
I’m as sure of it as I could be without actually seeing the source code. You don’t think they’re going to use Windows for the on-board navigation software, do you? No, those will be embedded Linux top to bottom – and that means GPSD watching the GPSes
>You don’t think they’re going to use Windows for the on-board navigation software, do you?
No. I expect them to use DOS. More precisely, GlaDOS. :P
@esr
> You don’t think they’re going to use Windows for the on-board navigation software, do you?
Maybe… The drones’ purpose is killing people, and you do know what BSOD means, right? I think the missions are quite compatible….
“Are you still there?”
“I seeeee yoooou…”
Barack’s hunter-killer robot assassin squad is generating enough controversy and negative PR when they’re killing the intended targets. The DoD probably doesn’t want to risk the collateral damage that would result from Windows crashing the things mid-mission. Especially given their experience with Windows rendering entire Navy vessels dead in the water.
Not Windows, no — but I can totally see them using VxWorks or QNX, especially for an aerial drone. Linux is nice, and it’s turned out to be a better fit for more niches than even Linus intended. But I still think even today there are easier and more appropriate approaches than Linux to do hard real-time stuff like drone control.
Those submarines don’t quite count as hard real-time; they cruise underwater at speeds less than ten knots, have a paucity of obstacles to avoid or circumnavigate, and most failures can be recovered from by tailcone kill, whether by an automatic safety or manual override from topside, and simply waiting for the dead vehicle to bob to the surface.
That said, when I worked at the submarine company we had one electrical engineer on staff who swore by Windows CE and thought we were wasting our time with Linux. :)
>I can totally see them using VxWorks or QNX, especially for an aerial drone.
If it’s QNX, they’ll be using GPSD; the QNX API is POSIX enough that porting GPSD would be the least-effort path. I don’t know about VxWorks.
Jeff: “I know, to a degree that many Americans let alone foreigners don’t, how shitty Obamacare really is in practice.”
As hard-left as you are, I’m stunned you’re not defending the shit sandwich as hard as you can. That says more about how Obamacare will work in practice than many, many other commentators’ efforts.
@Jay:
>As hard-left as you are, I’m stunned you’re not defending the shit sandwich as hard as you can. That says more about how Obamacare will work in practice than many, many other commentators’ efforts.
Hard-leftness is actually an indicator for being opposed to Obamacare, which in many ways is more centrist than anything.
In fact, a fair amount of the stuff that the left and right accuse of being each other’s evil plots are actually very centrist. Compromises often end up being the worst of both worlds, sometimes actively helped in that direction by monopolistic business interests that push for left or right wing policies according to what serves their purposes.
@Jay Maynard
> As hard-left as you are, I’m stunned
I think that is unfair on Jeff. I frequently don’t agree with him but he does not seem an ideologue to me at all. He seems to come up with his own conclusions based on the facts as he sees them, seen through a lens that I think is a little off kilter, but I certainly don’t think he is a parrot for a political point of view.
Having said that… I think his point is “Obamacare sucks, we should TOTALLY nationalize healthcare” rather than “Obamacare sucks, we should totally deregulated healthcare.”
Anyone with their eyes open is already seeing how badly Obamacare sucks, and how utterly dishonest the process and dialog to foist it on us truly was. I can’t believe they have the chutzpah to hold their head up in face of the whole “you can keep your doctor and keep your current coverage” debacle. But Obama, Pelosi and Reid were never short of chutzpah.
@Jessica:
>I think that is unfair on Jeff. I frequently don’t agree with him but he does not seem an ideologue to me at all. He seems to come up with his own conclusions based on the facts as he sees them, seen through a lens that I think is a little off kilter, but I certainly don’t think he is a parrot for a political point of view.
I’m not sure “hard left” means ideologue or that “ideologue” means parrot. One can be an ideologue for one’s own ideology, and one can hold views firmly away from center on an issue without being vocal about it. I’m not sure I’d call Jeff “hard-left” (certainly “solid left”, one line of evidence for that being that if he were any closer to center he’d be much more likely to endorse Obamacare), but being hard left wouldn’t automatically make him an ideologue, nor would call-it-as-I-see-it-ism preclude him from being one (actually, I think it would make him *more* likely to be an ideologue).
The hard lefties are the ones calling out Obama on his BS — including the half-assed sop to big insurance that is Obamacare. Read Arrow’s 1963 paper again — health care is not a market good. The solution is single payer.
The hopey changey Democrats you hear about on the evening news, still cheering for Barack even though once he got into office he championed every war, every spy program, and every government bailout to Wall Street he campaigned against in 2008, those guys aren’t leftists at all. They’re lemmings.
@Jay Maynard
> As hard-left as you are, I’m stunned
Jeff Read a hard-left? You must be joking. You obviously do not come out in the real world often.
What I hear about the ACA seems to indicate that the warring factions in the USA have managed to recreate Frankenstein’s monster in public services.
The idiotic thing is, that by just crossing the border (North or South), the US can see systems that work better and are much cheaper.
Winter: By US standards, Jeff is hard left. The standards of Europe don’t apply to US political discourse, nor should they. If I wanted Europe, I know where to find it.
As for crossing the border to the south to find a system that works better…what are you smoking and where can I get some? And, for that matter, Canada’s system isn’t working all that well if it has to have death panels…
@Jay Maynard
“Winter: By US standards, Jeff is hard left. The standards of Europe don’t apply to US political discourse, nor should they.”
These are more global standards than European. But I think the word I learned for your position was “Parochial”.
@Jay Maynard
“And, for that matter, Canada’s system isn’t working all that well if it has to have death panels…”
As far as I could determine, the “death panel” is an institute that decides when continuing medical treatment is unhelpful after the patient has become comatose and there is a conflict between the medical care givers and family about whether to proceed with treatment. If you think that is a bad idea, you obviously have not observed terminal care situations. Keeping a body alive after most parts of the brain have stopped functioning is only rarely humane.
I know that you want to insinuate that these death panels kill patients for money. But you need good evidence for that. Several US politicians accused the Netherlands of murdering patients for money because we have legalized euthanasia. That was all FUD and propaganda. So I assume the Canadian case too is just political FUD.
@Jay Maynard
“As for crossing the border to the south to find a system that works better…”
Mexico is a poor country. They do a lot to make the little they health care they have available to all. Their medical training seems to be good enough that Mexican doctors can work in the USA. That the USA with all their wealth are unable to even replicate the public health care system of poor Mexico is revealing.
@Jay Maynard
“what are you smoking and where can I get some?”
Personally, I do not smoke. But you can get such stuff in Amsterdam. Quite close to the Central Station. The easiest way is to leave Central station and follow the tourists to the left (East) into the red light district next to the station. You will find ample “Coffee shops” that sell high quality stuff. Do not fear, it is one of the safest areas in town.
https://maps.google.nl/?ie=UTF8&ll=52.374139,4.899312&spn=0.005423,0.014259&t=m&z=17&vpsrc=6
@Jay Maynard
I think the piece linked below sums up my thoughts on the health care debate quite well. It is also much closer to the way health care reforms are discussed outside of the USA. To us, the debate inside the USA seems to be happening between blind and the deaf illiterates.
Meet the Real Death Panels
Should geezers like me give up life-prolonging treatments to cut health care costs?
http://www.motherjones.com/politics/2010/07/health-care-rationing-death-panels
Winter, the health-care “debate” is like the global-warming “debate”: nonexistent everywhere except for America, where well-funded well-poisoners are given a voice and the opportunity to mislead.
And yes, by American standards I am far left. So is anyone sane.
“And yes, by American standards I am far left. So is anyone sane.”
Well, you’re 1 for 2.
@Jay Maynard
And how is it going with these infamous death panels?
Ask the Canadians.
http://www.powerlineblog.com/archives/2013/10/death-panels-yeah-thats-the-ticket.php
@Jay Maynard
That is referencing the same article I linked to. And that article is very clear about the fact that these Canadian panels are not about rationing at all. Instead they have to settle disputes on end-of-life question for comatoze patients.
Anyhow, you do not seem to have asked the Canadians themselves. Maybe you should read some genuine Canadian sources. It is not that there is a significant language barrier.
@Jeff Read
> And yes, by American standards I am far left. So is anyone sane.
Do you think that I am insane? Unfortunately, I hear this sort of thing from the left a lot. To me the assessment of insanity on the part of your intellectual opponents is more a reflection on you than me. To me it is an indicator of intellectual incuriosity, and parochial thinking.
I don’t think you are insane, though I do think you are wrong about stuff. Nonetheless, I am curious to understand how you come to your conclusions, wrong or right. I find it a broadening experience.
Labeling insane that with which you disagree with advances no argument except one’s own self puffery.
But, Fluffy, to the Left, the absolute rightness of their positions is so self-evident that you’d have to be insane to not see and agree with it.
@Jay Maynard
> to the Left, the absolute rightness of their positions is so self-evident that you’d have to be insane to not see and agree with it.
Unfortunately, the right does that too, as does pretty much everyone on opposite sides of any argument. This is supposed to be a place where the evidence speaks, and rationality reigns supreme, not a place for cheap shots. Such a blessed place is rare indeed, and I am shocked that someone as capable as Jeff Read would want to pee in the pool…
I’m not sure conservatives throw about accusations of insanity with the same wild abandon as the Left does. Ignorance, stupidity, blindness to all but emotions, yes, but not insanity.
@Jay Maynard
> I’m not sure conservatives throw about accusations of insanity
Counterpoint…
http://www.amazon.com/Democrats-Had-Brains-Theyd-Republicans/dp/0307353451/ref=la_B001H6GJTW_1_8?s=books&ie=UTF8&qid=1383516306&sr=1-8
That’s accusations of stupidity, not insanity.
@Jay Maynard
Not only accusations of insanity.
I have encountered people from the right who accuse, say, Canadians and the Dutch, to murder their old people to economize. Without taking an effort to think about the evidence.
But I agree with fluffy that these are characteristics of humans, not of specific political factions.