I am 67. I work in a niche field (Colour calibration / measurement). I work on projects, but I also have my own R&D software which has been growing for the last 20 years or so. I write non-fancy C++ most of the time. If you can find a small-scale niche like this, possibly involving hardware, then it will have longer timescales.

I have recently made a phone app. If you can find something that needs doing on a phone, this is something that one person can do. Android has a good community, but huge variations in hardware. Apple is, well, Apple, but their builds are consistent.

I spent a year or more on AI. The small successes I had were easily overtaken by the others in the field. Everything is changing furiously. It usually requires huge databases and a hefty processor.

Good luck!

I bet the patents will not stand. I worked for Canon in Guilford about 1990. They were working on 3D scanning as there was a bottleneck for getting 3D objects into the virtual world. One of the anticipated uses was for virtual shopping. Canon were keen on patents, so look for prior art about then. This was just before software patents, so you may have to search for a physical object, such as a 3D scanner or the mat with registration marks so we could identify the camera position from uncalibrated stills.

AI is a new tool. I am working on AI with image processing. Sometimes it works, sometimes it doesn't. So, do not say "This must be a fake, because Computers plus Science!'. However, it is very hard to quantify a general hunch that "It sorta doesn't look like most other Rubens stuff", particularly when other experts disagree. But if an AI can reproduce this feeling of a hunch from a scan of any small region, then it isn't just "the arm looks wrong" which may mean something was repositioned late in the process, but something generally wrong with every small bit of it.

The Van Meegren fakes of Vermeer may have had his style, but X-Ray studies showed he used titanium white instead of lead and zinc white, and stuff like that. The AI evidence is much less clear-cut than that. Presumably all the materials are kosher with this one. But there is something rum about it.

Disclaimer: I am not a Rubens fan. The picture is what it is. The real concern is the monster price tag because it is said to be by Rubens.

There was no-one available to take your package at *Mt. Doom, Gorgoroth* so we have left it with a neighbour at *Barad-Dur*.

My brother has a folder on his computer called OldComputr. This has an image of the disk on his previous computer. That has an image of the disk of the computer before that. all they way back to the first computer he owned, and then it contained files from other computers. As long as storage gets cheaper per byte this is likely to continue.

In https://web.archive.org/ there lies the Wayback machine. This does not archive everything. It does not archive everything that people push at it, so their version can be found and yours can't. But it does make an attempt to back up the more important bits of the web. If we want internet permanence (and people are posting for the right to be forgotten, too) then this might be a solution.

Railguns and their relatives are even older than that. The French were looking at magnetically propelled weapons them before WW1. I worked on pulsed power, though not on railguns, and someone how had worked on them described them as "a gun that fires its own breech twelve feet into the ground when there is anyone over the rank of Major watching". This doesn't mean they can't ever work, but the hundred-plus years of not working is a pretty strong hint.

Passing a milestone is a success. Hitting one not so much. I wish them success, but I suspect the text is accurate as it is.

This sounds about right. I remember it was estimated in the 1980's that a Cray XMP made an undetected error every 2 minutes or so because of cosmic rays. Cosmic rays are a particularly tricky thing to guard against. With steady, random errors you can add extra parity bits and trap when things go wrong, but cosmic rays give you a shower of ionising particles at the same time, and roughly in the same place, so there is a significant chance that it can flip a bit and the corresponding parity bit. Back then, when we were running hydrodynamic codes for many hours, the best thing to do was to check the mass and energy sums to see whether any significant error had occurred, and live with the minor differences. If something is really important, you do it twice and see whether you get exactly the same answer.

Imagine you have the choice of two cars. One has been assembled by top engineers to the highest standard but they haven't started the engine or pumped the brakes. They may have flicked on the indicators and the lights, and checked the instruments, but that's it. I have been using the other one all week and it worked fine. Which would you chose?

Being human-rated is a good declaration of intent. It says 'we have worked on this booster as though it was going to carry people'. But it does not make the rocket safe. But the re-use will.

Reasonable question. Here goes..

Ptolomy approximated the motion of the planets using epicycles: toothed gears with integer ratios. These are good at producing the looping (retrograde) motion of planets such as Mars with a very simple mechanism. There is no evidence he (they) believed the planets ran on wheels - these were for calculation only.He is supposed to have used 48 epicycles, with later modifications to over 50 but probably not as high as the figure of 80 that is sometimes quoted. Simpler models incorporating some of the same ratios date back to the Babylonians.

No-one knows of an earlier machine, but it is entirely sensible that Ptolomy's calculation was backed by a physical machine. Egypt was conquered by Alexander the Great, who was in turn conquered by Rome. Archimedes lived with other Greeks and Egyptians in Syracuse, which was in Magna Graecia (the bottom end of Italy and Sicily that still kept a lot of the greek character at the time). It is unproven but perfectly reasonable Archimedes might make a compact version of the Ptolemaic mechanism for a Greek shipping magnate for calculating dates and tides. This would be valuable, so was probably kept secret, and the device disappeared from recorded history.

We start from a device that has been carefully made compact. We would expect at least 48 gears from Ptolomy's gear train. We see a number of gears with the right Ptolemaic ratios in the right places, plus some sophistications, so we are probably looking for 50-plus gears, plus perhaps a few extra to deliver the calculations to the display. The bits we have also suggest that it fitted into a rectangular box, which also confirms our guess of the overall size and complexity, and fits with some sensible guesses on how the front might have looked.

Please?

I have occasionally seen people with amazing qualifications on paper who turn out to be useless, and people with no qualifications who are great. The solution at Canon Europe was to have a simple programming test, and to write and give a presentation. The test was simple - write (say) a simple sort algorithm, what is the prescience of these operators, how would you support different image formats? Off the wall answers were encouraged. I remember answering the precedence of operators question with (1) I know the page in K&R that has the table, (2) no-one should write code like that, and (3) the answer (which may or may not have been right but it hardly mattered).

If you know the basics, it gets the interview ball rolling. And if you have any large gaps, they will show. One person applying for a senior sysadmin post went to the toilet during the test and never returned.

I see two sorts of programmers. One can manage a small program in detail. It can be tens of thousands lines long, but not millions. I am one of these; you may be too. Then there are others who can happily stitch together lots of different pieces of code, and make something huge out of this. I sometimes can't even read their code: it is all classes and wrappers and producers and factories, and hard to find what actually does something. Or, you have the data you want *over here* but you can't see how to get it *over there*.

I am a small-scope coder, so I like to imagine a big program ( LibreOffice for example) as a collection of small programs I can understand. If I wanted a feature, such as importing a new image format, then I could write a little, sandboxed app. This could work without code review. If my became popular, it may be necessary to code review it, and to incorporate it in the big program so it could run more slickly. A big-scope coder might see this differently. They would see the whole project as a structure of classes and inheritances, and having independent rogue bits is nasty and baad. And they may have their points.

Life on Earth had several critical stages...

Everything seems to have a common genetic code, though there are other ammonia acids that cod have worked in DNA: so, this probably evolved once.

The evolution of the Eukyrotes is believed to be some to be the merging of two bacterial forms giving a doubled cell wall. This makes them more robust and more efficient. All multicellular life seems to come from these.

It is guessed that there may have been an earlier form before the known Archaea. It is possible that life could have started off as something much slower, more akin to crystallisation. It is quite possible that life did not evolve from nothing in thermal vents, but there was a precursor, which would probably have all been eaten by the newer, faster lifeforms.

A lot of evolution is a slow selection of the fitter forms. A lot of the sophistication we now see on Earth is the result of this. But there were some significant hurdles to overcome. To my mind, the existence of perchlorates in Martian soil is evidence that there is no life. Not because perchlorates kill germs: they are energetic molecules, and if there was life, it would have found a way to eat it.

This will be O(N) but it will be slow for a million data values. With 2^32 values to set to zero, your array is about 4000 times a large as your data. It would probably be overtaken by an efficient O(Nlog(N)) scheme even if you had an efficient way of zeroing your array in blocks.