The few thousand copies is over the _entire_ lifetime of a book, not meant per year. Again, your College 101 book in Physics or Calculus will be used in large courses and thus sell a large volume, if it is being that used that is. There are also many competitors in the market, and yours might not see significant pickup.
So, let's suppose that you indeed sell 1,000 copies at $150 a piece. Your share is 6% of the net sales, so $8370. From this you need to pay services such as the person making the index, or pay royalty for pictures (yes, many publishing houses deduct this from the author's earnings).
Let's say your book has 700 pages, which is pretty normal volume size, and after you paid for additional services and expenses you are left with $7000. You are a good writer, so planning the book, writing it, creating examples, copy-editing and rewriting it, making reference solutions, etc. and you only spend on average 4h per page. You have made $2.50 / hour before tax. And you probably have a PhD in that topic.
No, almost nothing of the $200 actually goes to the author. A typical share goes like this: (concrete numbers for Germany) 50% of the sticker price goes to the book seller, 7% to tax, the remainder to printing, typesetting, logistics, the publishing house etc.
A typical author would get somewhere between 4-7% of the net sales. Most books, especially advanced science books, on the market (not your College 101, not NYT bestsellers) sell only in the order of hundreds, maybe a few thousand copies. The vast majority of books doesn't even earn the advance back. The reason why they are often not good, is that it takes months to write 500 pages of well-thought out, well-delivered material, and the incentive to do that for $5000 is relatively low.
> So depending on the rejection rate, which I think is ~95% in the case of Nature, you're paying for those other 19 papers to be checked.
Yes, the other 95% need to be checked, but reviewers are recruited from academics and typically do this task for free. The contribution that the publisher does at the stage of manuscript evaluation is marginal.
On rapiscan models (backscatter scans) it is basically software controlled, although the modularization makes it non trivial to be changed in a hack, as reported in this talk: https://www.youtube.com/watch?v=x_f4HUrn-NA
With "special services" insufficiently defined, this opens up a lot of possibilities. Just today, Deutsche Telekom announced that
"According to our [Deutsche Telekom] ideas, they [startups] pay for it as part of a revenue sharing of a few percent. That would be a fair contribution for the use of infrastructure."
With "video conferencing, online gaming, telemedicine automated traffic control and self-steering cars up to integrated production processes in the industry" as examples of users needing to pay, this goes a lot further than what has always been argued for by the commission.
The committee voted against strict network neutrality, and for the possibility of differentiated "specialized services". The bill now goes to the European Parliament for vote.
If you're flying from Amsterdam to the US, you do. They have usually blocked off the metal detectors and claim that you have to use the body scanners (you do have the right to opt out, but it's not stated there).
From their website, it seems that tarsnap can't be counted as OSS:
"The Tarsnap client code is built around the open source libarchive archive handling library. While the Tarsnap code is not distributed under an open source license..."