Posts by Francis Vaughan

I would say that the NBN are in the mode of under-promising and getting ready for over-delivering right now. Looking at the current maps for in-build here in SA there are huge swaths of FTTC in build that are slated to start to be available in May - areas where there the ducts and pits are right now installed ready to go. Further massive amounts of FTTN slated for October - many places where there are already node boxes installed, ducted and powered, ready in the streets.

I am suspicious that the NBN is managing expectations versus delivery in a way that will start to make them look really good in a politically useful time-frame. If a large fraction of the populace suddenly start getting FTTC with really good performance this year, much political capital will be made by all involved.

I was one who was rather peeved about the failed app. Especially as I live in a high risk bushfire area.

The app did used to be pretty reasonable. Although I thought the GUI design was actually rather B grade (the icons used to denote the events made very poor use of space and were difficult to distinguish.) I was rather surprised to discover that the whole system is little more than a data scraper and a server that pushes the scraped data out to the app. The alert data is scraped from various government sources and aggregated. That is it.

The web site for the company gives you the sinking feeling that it is a tiny one or two person startup. They probably pitched the app to SA and Victorian state governments and asked $250,000 a year for it. If they had made it stick they would have been on to a good little earner. No doubt hoping to get the other states in as well in time, getting to the point where they might be pulling in say $3 mill a year. And really, not for much. But, as noted above, the update to the app last year made it perform worse - they lost my user settings when they did it, and it seems they simply never had the server resources set up or the required resiliency in the system to justify the price they asked. Altogether a bit sad, but also a very common story.

Seriously, the vast majority of the comments clearly have no clue as to what the device is, or how it works. Everyone seems to be thinking of tech they read about decades ago. First up, look at the name of the technology. "Digital radio frequency memory". There is a clue here.

Radar works by the missile sending a pulse out, and listening to the return. In modern systems that pulse shape can be very smart, and radars can see a lot about the target from the pulse return. Not just signal strength, but doppler shift and a host of second order clues. The point of a DRFM system is to spoof the missiles radar, knowing that this is the sort of thing it is going to be trying to do. So how? In principle it is easy. Receive the pulse, record it (the "memory" bit of the name) slam it into a custom bit of seriously fast custom silicon or FPGA, and work out what the return pulse needs to look like to make the missile think what you would like it to. It isn't necessarily a matter of making the can look like your plane directly. The can can spoof the doppler profile expected from your plane. Even if the missile changes its pulse profile, the DFRM system will continue to work, as it records and replays each pulse as needed. This is not your grandmother's radar jammer.

As to optical missile and guidance systems. Active laser systems are already available with the ability to blind incoming missiles. There will always be a mix of radio and optical threats, the presence of one does not mean you discard managing the other.

It is exactly an arms race. Missiles can have upgraded software to try to work past DFRM systems, and those same DFRM systems can have upgraded software to cope with that.

As described it was never going to be a product. That was just a mish mash of stuff with no clear use case and more importantly - no software to take advantage of it.

IMHO the big problem was that the processors had no sensible architectural support for making use of a world with huge addressable persistent memory. My personal crusade is for tagged memory. HP are one of the few companies that could conceivably create a commercially useful architecture with this. IBM being the obvious other. Tags for data type, if a pointer, access protection, and concurrency control, at a minimum would make life vastly more interesting. You can tie concurrently control and pointer identification into your memory network control. Suddenly lots of optimisations are available at a hardware level, and you eliminate a whole raft of crud from software.

None of this is exactly new. IBM's AS 400 was on the way, and it was a commercial success. And there were many other small volume and research systems built. But the ubiquity of x86, Windows, and Linux ensures that the barrier to entry of a properly new paradigm is very high.

Not noise cancellation. Not at all.

As noted above. This isn't noise cancellation. Noise cancellation on open spaces is astonishingly difficult, and it only works in headphones because the distances involved and the acoustic environment is so small. (And why the heck does everyone think Bose invented this either?) Once you have a reverberant field you are pretty much sunk.

You have been able to buy noise blanketing systems forever. Before they were electronic, a wire brush rotating inside a metal cylinder made for a very effective white noise generator.

All this offering is is an expensive way of making the noise source an IoT device that looks cool. There is no real technical innovation from the actual acoustics point of view over something you could buy 50 years ago.

Lunacy.

This is going to be a self inflicted wound of mortal proportions.

The absolute last thing any purveyor of ads ever wants its customers to know, is how much that customer's adverts did for them. Because the answer is - nowhere near as much as the money spent would justify. Ad effectiveness is the big secret. Letting that particular cat out of the bag could ruin Facebook and Google overnight.

This is probably a good thing.

More likely, the system will quietly be decommissioned when the first round of deeply embarrassing results start to appear.

Spin

Lordy, what a crock this article is. As noted above, Apple talk about first owners, not product lifetimes. And more importantly, this is information about environmental impact. Apple are being hard on themselves here. It would be in their interests to suggest longer times, and thus improve the environmental rating of their products. Had they suggested significant;y longer times they could (and should) have been taken to task for trying to soft pedal the impact of manufacturing their products.

As any owner knows, Apple stuff tends to be very well built, and outlasts kit from just about anyone else. Further, it keep its value way longer than other kit. You can argue why it might, but the reality is that it does. As products that are viewed from the standpoint of environmental footprint Apple do very well.

Tsiolkovsky rocket equation

The value in cooling the fuels is in the Tsiolkovsky rocket equation. The delta velocity of a rocket is proportional to the natural log of the ratio of total mass to dry mass. Anything at all that adds to the dry mass rips performance from the system mercilessly. Eventually every kilogram of dry mass added is a kilogram of payload not reaching orbit (or at least staging). Just making booster tanks bigger does not yield performance improvements as well as one might wish. Adding more fuel (and hence total mass) without any penalty in dry mass is a bigger win than one might expect. The higher effective fuel flow possible can translate to higher trust - which helps with that additional total mass on liftoff, something that matters a lot when most of the thrust is simply vanishing into balancing gravity. An even marginal change in thrust can have big knock on effects. Overall the ultra cold fuel is as close to a free lunch as one might wish for.

C is a glorified assembler

The trouble with C is that it is little more than a glorified assembler. The language lacks a significant number of structures that we take for granted in more modern languages, structures that make life easier when building more complex code.

But you can convince it to do things that are useful. Often with gruesome distorted bits of code that at first sight make little sense. The problem isn't that you do this, it is if these bits of code are not rigorously codified and documented, and wrapped up in a clear and maintainable manner. Macros can be one way of doing do (with a lot of care). But clear and enforced coding rules for their use is the key. There should never be a time when anyone looking at the code is in any doubt about what is being done and why. If the idiom is useful, I would expect it to be used everywhere as the standard mechanism for such condition handling. Personal idiosyncratic use of such features is the problem, and speaks of deep problems in the project.

The failure in the SSL code is not the use of the code, but rather that it is blandly written with no apparent consistency with the rest of the source. Just a one time use of an idiom. That suggests almost zero care, and is deeply worrying as a pointer to the quality of the rest of the code.

LOX

The problem with the latest Falcon 9 and its fuels is that they are now using a much colder than normal fuel. Both the LOX and RP-2 are chilled down significantly colder than other boosters. That allows the same tank structure to hold more fuel, and thus extend lift capability. Quite significantly it turns out. The down side is that keeping the fuel cold is a much more critical and difficult task than it has been in the past. They have to load it much later in the countdown, and there is much less wiggle room with delays.

Logs

The other sad truth is that both the clinicians and police seem to be utterly naive. How do they think they were caught? In an age where the popular press is full of how your every move can be tracked with your phone, exactly what logs do they think might be kept about their accessing of confidential records? Logs that won't go away, and can be reviewed way after the event if there is the need.

F16

For all the hand wringing about MMH, one should remember that the venerable F-16 powers its APU with MMH. The ground crew need to refuel the planes if there has been an in-flight problem where the APU is needed. And accidents do happen refuelling. Apparently MMH bleaches shirts and stings eyes and skin. Then results is a week of daily hospital checkups.

Fair Go

Who doesn't remember sketching such inventions in their exercise books during a really dull afternoon at school? Sure, once you reach puberty idle thoughts turn to more base topics. But as a seven year old such inventions were part of life. I mean, there was this fabulous documentary on TV all about such ideas, let me think, yes, it was called Thunderbirds.

Scale

One of the problems with criticisms of the design of the car systems is that it doesn't fit the mindset of the car engineers, and places a model over the car that actually doesn't exist in computers either.

Last I saw your average PC was just about as open a trainwreck as the cars we are criticising. There are a huge number of separate processors, many interconnection buses, and zero security. A PC typically has a number of high speed buses (SATA for a start) talking to subsystems with their own embedded operating systems. Then there are the slow buses for trivial stuff (USB 1.1 devices) and faster USB for things like WiFi and Bluetooth. Every one of these device controllers has embedded processors, many with subvertable hardware, and known attack vectors.

I don't hear pious moaning that it should be trivial to add firewalls to all the buses inside a PC. Yet it is essentially the same problem. There are hacks that can pwn a hard disk drive (many of which run say three separate ARM processors and a full multitasking OS). Not to mention hacks that can subvert your ethernet controller or WiFi controller to take over your PC. We all know not to plug an unknown USB device into a PC - but I bet that is a rule more observed in the breach. It isn't trivial autorun exploits we have to defend against now.

Yes, car system security is a big deal. But don't pretend that somehow the mainstream computer industry has trod these tracks long ago and it is the car engineers that are dolts. Everything is built to a price, and when there isn't a clear driver for change, change doesn't happen.

The care taken in car system where the issues are understood makes the mainstream computer industry look like a bunch of idiots blindly walking into walls. These are hard real time systems, and they are tested and simulated to clock edge and instruction boundary precision. But like so many stories of security in the history of computing, nobody even thought it was an issue. (Like the Morris worm, when the first message that went out had the point that was along the lines of - "we all knew this was possible, we just didn't think anyone would be stupid enough to do it.")

Energy

The article pretty much fails to take account of energy.

Given an arbitrary lump of mater floating about in space - say some planetoid, there is only going to be a limited amount of recoverable useful energy resource available for it. That new energy is what will limit the number of useful droids the exponential manufacturing can create. However you need to subtract the energy needed to mine that energy. Unless you have some form of matter converter to create energy directly from matter. The existence of which would be a game changer that would make droid armies the least of your plot worries.

Also, an asteroid travelling at speed is identically efficient as an energy beam. Both carry energy. It is all well and good to point out that a large rock travelling at relativistic speeds carries insane energy. But how did it get to those speeds? The energy it carries has to be imparted to it from something, and the energy must balance. If your rock carries a 100 zillion joules of energy because it is moving at 0.9C you need technology that can usefully direct at least 100 zillion joules into it. Why not just deliver that 100 zillion joules directly? It will have the same effect.

"Graham calculates that cracking 1024-bit DH it the computational equivalent of 2.5 hours' worth of global Bitcoin mining power."

The article also stated that the NSA system took a year to crack a prime, and cost $100m

Thus the equivalent value of the global bitcoin mining equipment is $350 billion.

Hmmm. Obviously there is a lag in technology between the NSA system and now, but there remains something of a gap here.

Old style acounting, and old style plans

Who has a mobile phone plan where they actually end up paying for individual calls? Seems most telcos and resellers are selling plans that feature some idiotic number of calls "for free" in the plan and then simply use the call count as a way of pushing you up a tier in the plans if you actually make a lot of calls. Under almost all usual circumstances you never actually pay for metered calls.

Getting rid of premium numbers OTOH would be the proverbial good idea.