Hmm, interesting. It might be worth pointing out that Skype was originally based on a decentralized service pushed through the Kazaa network:

http://arxiv.org/abs/cs/041201...

Like its file sharing predecessor KaZaa, Skype is an overlay peer-to-peer network. There are two types of nodes in this overlay network, ordinary hosts and super nodes (SN). An ordinary host is a Skype application that can be used to place voice calls and send text messages. A super node is an ordinary host’s end-point on the Skype network

Of course, the problem with the Skype system (as it was when that paper was written) is that the decentralised nature of the network means that your video call could be routed through any number of Skype network nodes (i.e. computers) before it arrives at its destination. I think now Microsoft has replaced most of the supernodes with microsoft servers, so replace "any number of Skype network nodes" with "any number of Microsoft servers".

Presumably Tox is doing something similar to going back to the roots of Skype, with maybe a bit more encryption thrown in.

Then I thought, well perhaps designer spends years designing a game with all sorts of clever ideas then copiers use them all a few days after release. I have to ask, though, is this what happens? Surely a game must spend some time before becoming popular enough to copy, during which it builds a following and has first mover advantage.

Flappy bird is certainly not a good example of the ideas being the expensive part. Here's just one example of an earlier game that is similar in nature:

http://www2.sunflat.net/en/gam...

Right. I finally got around to writing an R function to do this, because this problem has cropped up a few times in the past year:

getPV <- function(prevalence, sensitivity, specificity){
        popnTrue <- prevalence;
        popnFalse <- (1-prevalence);
        popnTruePos <- popnTrue * sensitivity;
        popnFalsePos <- popnFalse * (1 - specificity);
        popnTrueNeg <- popnTrue * (1 - sensitivity);
        popnFalseNeg <- popnFalse * specificity;
        ppv <- popnTruePos / (popnTruePos + popnFalsePos);
        npv <- popnFalseNeg / (popnTrueNeg + popnFalseNeg);
        return(data.frame(prev = prevalence, sens = sensitivity,
                                            spec = specificity, ppv = ppv, npv = npv));
}

NCI tells me that 4% of the US population are cancer survivors, so I'll use that value for the population prevalence:

> prev <- 4 * 0.01;
> sensSpec <- rbind(c(94.8,54.7),c(81,78.7),c(62.1,94)) * 0.01;

> out.df <- NULL;
> for(i in seq_len(dim(sensSpec)[1])){
    out.df <- rbind(out.df,getPV(prev, sensSpec[i,1], sensSpec[i,2]));
  }
> out.df;
    prev sens spec ppv npv
1 0.04 0.948 0.547 0.08020305 0.9960546
2 0.04 0.810 0.787 0.13677812 0.9900409
3 0.04 0.621 0.940 0.30131004 0.9834779

So the best they can do for this test, according to the paper, is a 30% positive predictive value -- if this test comes up positive, there's a 30% chance that you actually have cancer (and that's allowing for 2% of "negative" results actually being cancer).

The actual paper is behind a paywall.

Yay for institute access. Their idea of "approach[ing] 100%" is a little bit loose:

Based on these calculations, the cutoffs for low (0.10), medium (0.25), and high (0.50) thresholds are 1.47 at a sensitivity of 94.8% and a specificity of 54.7%, 1.73 at a sensitivity of 81% and a specificity of 78.7%, and 1.99 at a sensitivity of 62.1% and a specificity of 94%, respectively

I have yet to do the calculations using population prevalence, but I'm going to guess that the positive predictive value of these tests are not particularly high.

For that matter, all of everything constructed by human beings

You might not be terribly surprised to know that our genes (and the genomes of pretty much everything) are also full of bugs. We have a whole raft of deleterious genetic variants in our DNA that are just waiting for the perfect time to activate and say "hey, you know that life thing? I can make it worse." On top of that, we have a few viral genomes in our DNA (possibly some that are still active), and rely on bacteria and mitochondria to provide us with energy required to live.

In other words, defective objects are the rule, not the exception.

p.s. hmm... I've only just realised how much I miss that handy login form that SoylentNews has to deal with accidental AC posts.

We apologise for the inadequacies of our car at high speeds, and are investigating ways to make it even safer. We have designed a flexible partitioning system to take some of the energy from a "car split" incident, and will be implementing it in all new Tesla cars, and retrofitting it to all drivers who want it. Additionally, the car will require that the driver and all passengers are wearing seatbelts when the car is driving at speeds exceeding 70 mph.

My wife just pointed me at the Thermomix, which is popular among her German friends:

http://www.amazon.com/Vorwerk-...

It can weigh things, grind grains, and chop, as well as all the other kenwoody-things. It's a bit more expensive, though, and probably not induction.

Microsoft notes that it worked with multiple international companies to secure its version of the standard.

Ah, yes. Once again, Microsoft has their own special idea about how to extend a standard. Said like a true Microsoft employee (or paraphrased by someone with a strong reporting bias -- it doesn't seem to be phrased in this way in the original Microsoft post about encryption and transparency).

SoylentNews has decided to avoid non-profit status due to the demands it puts on the organisation, so they're now trying to set up as a slightly more normal "we don't actually want to make money" benefit corporation.

An induction cooktop with precise digital temperature control (SI) and a magnetic stirrer would also be great.

A magnetic stirrer on a magnetic induction cooktop would be... interesting.

We have an induction coooktop with digital temperature control (in increments of 10 degrees). It seems to measure the temperature at the induction coil, rather than the temperature of the pot, so things can boil when it's set to 60C. Also, the PWM cycle of the cooktop (as with pretty much every other one I've seen) is far too long at about 0.5Hz (where I'd prefer a cycle of at least 10Hz, and ideally over 100Hz). Further, the power level can't be adjusted as much as I'd like -- I set it to 800W (or 130C, because that seems to be similar) and it's too cold for frying, but 900W (or 140C) is a little bit too hot.

Sure, I wouldn't change away from induction now that I have it, but I expect it'll be a while before we get a replacement cooktop, because I've become a whole lot more aware of the limitations (and possibilities) in the current technology.

And now that Tesla has freed up the patents for their superchargers, you'll be able to plug an electric bike into something that uses that connection and current (not necessarily the Tesla ones). Given that the motorcycle battery packs are much smaller than the car packs, I don't expect that a 2-minute charge to full would be out of the question.

That might almost be quicker than walking up to a cashier and paying money, and certainly would be quicker if you're not the first person in line.

And, of course, they can snoop on American citizens on google and facebook, as well as for all other communications in Great Britain because the Americans are foreigners.

When you have five eyes, and each eye is in a different country, it's quite easy to work around those pesky "no watching yourself" laws.

It's extremely confusing to put it nicely.

I feel compelled to tell the world about a more confusing part of NCBI that I'm trying to navigate myself around at the moment: The Transcriptome Shotgun Assembly Sequence Database. Submitting sequences is... a little tricky. Here's a simplification of the process:

1. Create a BioSample record for the organism that you're submitting data for
2. Download a sample template tab-delimited file, and fill in arbitrary descriptions about your organism
3. Upload the template file (using your web browser), and finish the remainder of the BioSample submission process
4. wait for email confirmation of your BioSample record, after which it will have an "official" ID
5. Create a BioProject record for your transcriptome assembly project, and link in the BioSample record (I don't think you need to wait for email confirmation to get an ID for that)
6. Create a Sequence Read Archive (SRA) record for your transcriptome assembly project
7. Create an experiment record (in the SRA record) for your transcriptome assembly project, one for each different method of sequencing that was used
8. Get md5 sums of all the raw data files that will be uploaded to NCBI
9. Create a run record (in the experiment record), and add in the file names and md5 sums of the raw data files
10. Upload your files to the NCBI servers using an FTP client
11. Wait for files to be transferred from the NCBI FTP server to the SRA server, after which the run record will get an official run ID
12. Create a Transcriptome Shotgun Assembly (TSA) record for your transcriptome assembly project, and link in the BioProject and BioSample records, as well as the run IDs from the SRA record
13. Use a web form to create a metadata file to download to your computer
14. Use a custom NCBI program to merge the metadata file with your transcriptome assembly
15. Upload the [large] merged file to NCBI using your web browser
16. Wait for email confirmation, after which the TSA record will get an official ID

Congratulations, you are now the proud owner of a Transcriptome assembly ID, which you can insert into a single sentence in your research paper: "The transcriptome that was created for use in this study has been uploaded to NCBI (reference ID: GAAA00000000)."

$10,000 barely gets you ONE modern well-equipped 20 core server system

I get 4x16 core AMD Opteron 6366HE on a Dell PowerEdge m915 for $5,578.70:

http://configure.us.dell.com/d...

So that's a bit less than $10,000 for 100 cores on a standard issue Dell machine. It's not completely crazy to expect you could increase that to 600 cores without too much extra cash laid down.

You need to click on the "Elsevier Open Access" link from NCBI, which is a direct link to the article on the publisher's website (this location is where you click for all PubMed articles, as long as the publisher has provided access in that way). PubMed never displays complete articles.

After clicking through, there's a "Download PDF" link at the top left of the article, just under the green Science Direct header.