All heart, no brain

I started watching The Waters of Mars, the most recent Doctor Who special, a few days ago. Then I paused it after 12 minutes and didn't resume for a few days, because I had other stuff to do and it clearly wasn't going to be very good.

I know Doctor Who is really fantasy, not sci-fi, and I know it's now all about heart and emotions and not so much about coherent storylines. That's fine, if done with some imagination; I actually quite liked the episode Gridlock, for instance, which was a veritable lace doily of plot-holes if you looked at it critically.

And I know Doctor Who is primarily aimed at young viewers, and I also know that kids aren't very discriminating and will watch any old crap.

But none of that excuses this level of crapness.

(Spoilers, naturally, follow. But I'm spoiling the bad bits, not the good ones, so perhaps you'll come out ahead.)

The Waters of Mars reminded me of Robert L. Forward's excellent (if you like hard sci-fi) Dragon's Egg (the sequel's pretty decent, too!). The only purpose of the characters in the first couple of dozen pages of Dragon's Egg is to set up the story proper, so Forward obviously didn't see any need to spend more than a lazy half-drunk afternoon writing the first part. (Actually, I wouldn't be surprised if he wrote the first part last, just to give the audience a minimal on-ramp to the highway he'd already finished and was impatient to publish.)

So Forward, for instance, details exactly the garments which a young female astronomer puts on before racing off to advance the plot, but because he's not paying attention he gives her a skirt but forgets to mention any underwear. This is forgotten once you get into the real story, but it's somewhat startling at the time.

Likewise, in The Waters of Mars, the writers are clearly so eager to get to the, super-heavy-handed but still pretty neat, ending and teaser for the upcoming Christmas special, that they just didn't care about the preceding story.

Robert Forward's dodgy beginning bit was very small. In The Waters of Mars, the dodgy beginning bit takes up five-sixths of the show.

I could just about handle Mars gravity being the same as Earth gravity, when it ought to be less than 0.4G, because that's apparently still too expensive for live-action TV to do properly. And I could barely accept explosion debris cheerfully burning away in Mars' 95.8%-carbon-dioxide, 0.2%-oxygen, less-than-1%-of-Earth-pressure atmosphere, because, um, maybe this Mars-base was built out of bamboo packed with potassium nitrate.

But the monsters are creatures that can make water (and fusion power!) out of nothing. But they're desperate to get to Earth, because there's so much water here. (And they've got the same name as the principal villains of all of the Halo games.)

All the writers would have had to do was make the monsters express a great hunger for all of the people there are on Earth for them to infect, or specifically mention how pleased they are with Earth's ever-shrinking ice-caps that promise a gigantic habitable area for them in their liquid form. But no. One of 'em stands there, drooling a steady stream of water onto the floor, and just says that it's impressed by the quantity of water that Earth already has.

Cliched self-destruct

And there's not just one, but two, self-destruct mechanisms activated in this one episode.

I suppose it's not that surprising that the systems exist - nobody puts a "Blow Up This Vehicle" button on real-world dashboards, but if you live in the land of TV sci-fi you can expect super-virulent body-snatching alien and/or supernatural monsters to pop up about every other week. The only surprising thing is how slow people always seem to be to figure out what's going on and press that deadly button that'll save the rest of the world.

(We should probably count ourselves lucky that only one of the self-destructs has a Red Digital Readout. And to be fair, it still isn't your typical Acme Mechanically-Assisted Plot-Tensioner, a device which has the mystical ability to make the last 60 seconds of the countdown take up five minutes of screen time.)

As regular readers know, I am actively delighted by stupid Doctor Who monsters. But they're meant to be stupid-looking, not just by-the-numbers Central Casting zombies plodding through a script that exists only to give the Doctor a reason to emote.

I'm quite happy with fatally-plot-holed sci-fi as long as it's imaginative. When I finish watching some oddball anime and say "what the fuck was that all about?!", I'm always smiling. And Doctor Who is supposed to be among the most imaginative live-action shows, because it's got the fewest restraints. It's not stuck on a particular starship or even a particular planet, it doesn't take itself very seriously, and after some decades, the audience is accustomed to the fact that the TARDIS seems to independently seek out deadly peril, especially when the Doctor intended to have a little holiday.

This all makes it particularly disappointing when you get a story like this, that's no better than the 62nd time the holodeck tried to kill everyone on the Enterprise.

Pew pew pew! ZAP! Whoosh! Ka-BOOM!

You know when you read a review of a game that says that one part of the game, say the battles between spaceships, looks great and is tons of fun, but the rest of the game is kind of boring?

Gratuitous Space Battles is that part of that game, without anything else.

(And before I say anything else, note that there's a free demo.)

You pick a fighter, frigate or cruiser hull for each of your vessels...

Gratuitous Space Battles ship design kit them out with weapons and shields and engines and so on, you deploy an armada of ships of different sorts (or all of the same sort, if you like), and then you give them all orders. Concentrate all fire, prefer to shoot enemies that're already wounded, shoot this kind of ship over that kind, protect this ship of ours, protect any ship of ours that's damaged, stop at this range from the enemy and plink with your long-range missiles rather than charging into beam range, et cetera et cetera.

And then you click the "Fight" button, and sit back and watch.

For the actual battle - which is fought on a 2D battlefield, though ships can go over and under each other - you're a pure spectator. GSB is like a tower defense game, in that regard. (Many tower-defense games let you build new towers during a battle, though; GSB does not.)

You can speed up and slow down the battle, and you can zoom in and out. From a distance, the action looks like this:

Gratuitous Space Battles wide view

(In this battle, I'm employing the Unsporting Crowd of Torpedo Frigates strategy. I'm also playing at full resolution on my huge monitor, so the full-sized screenshot is 2560 by 1600 pixels and rather a lot of kilobytes.)

Zoom in, and you can see...

Gratuitous Space Battles zoomed in

...each individual weapon shot, repair drones patching flaming holes in hulls, and fighters weaving around the capital ships. (Full-sized screenshot here.)

When you win a battle you earn "honor" with which to unlock new hulls, equipment and the three whole alien races besides the one you start with, the Federation. (The big Federation ships, rather delightfully, all look like a hybrid of a Starfleet vessel and a Battlestar.)

It's all a lot of fun, and should become even more fun as the game expands. Cliff Harris, the indie developer of GSB and a few other games, is actively patching bugs and adding stuff, and GSB is also very moddable. Fans have already, according to the ancient tradition of the first few mods for any game, created a few rough-and-ready super-battleships by just adding more module mounting points to existing hulls. Some proper high-quality mods with all-new graphics, like unto the Babylon Project mod for Weird Worlds, should be arriving soon.

So try the free demo and see what you think. The full game takes into account what you've done in the demo, by the way, so you won't have to play the tutorial level again if you don't want to, and get to keep whatever honor you earned.

(GSB is Windows-only at this point, but because it's not a very demanding game it generally works fine on other OSes if you play it in an emulator.)

Gratuitous Space Battles is $US22.99 from the developer, or only $US20.69 on Steam.

Note that there's a graphical glitch in GSB that affects people who're using an unusually high horizontal screen resolution (so, one giant monitor, or a row of smaller ones). It...

Gratuitous Space Battles screen glitch

...turns a column of screen to the right into stripey repeats of the last correctly-drawn column of pixels.

I think this was meant to be fixed in the recent patch, but it doesn't seem to have been. No problem, though; just go to the options and disable "Gratuitous Shaders", and with very little eye-candy reduction, the whole screen will draw properly again.

Know everything they type, or stop them from typing at all!

Back in 2000, I reviewed the KeyGhost Security Keyboard, an apparently ordinary keyboard with a hardware keylogger hidden inside it. Later that year, I reviewed the KeyGhost II Professional, another hardware keylogger, this time built into an innocuous-looking keyboard plug adapter.

Those reviews have a special place in my heart, partly because I just love the sneakiness of these little things, and partly because someone ripped off my pictures of the guts of the Security Keyboard...

Keyghost unit

Keyghost unit side 2

Keyghost unit side 1 create an urban legend about hardware keyloggers allegedly being built into Dell laptops. (Or other makes of computer - the story's had a few mutations over the years.) Some people appear to have decided that the fact that the pictures and info about the hardware are obviously copied from my review means that I'm part of the conspiracy.

(KeyGhost now offer a Mini PCI keylogger, which actually could be hidden in a laptop computer with a spare expansion slot, or in a desktop machine with a Mini-PCI-to-normal-PCI adapter. I'm pretty sure they're not selling them by the million to the Department of Homeland Security, though.)

Anyway, KeyGhost don't sell those exact products any more. They've got better ones. And a new gadget with a completely different purpose, whose value it took me a little while to see.

The old Security Keyboard I reviewed had a memory capacity of half a million keystrokes, before new keystrokes would start overwriting the oldest ones. The KeyGhost Pro had a compression system that let it fit rather more keystrokes into the same amount of memory. And they weren't particularly cheap; the Security Keyboard version I reviewed listed for $US299, and the KeyGhost II Professional was a $US249 item.

Nowadays, you can get a 128,000-keystroke plug-adapter "External KeyGhost Home Edition" for only $US89, and for the price of the old Security Keyboard you can get the KeyGhost Professional SE Security Keyboard, with more than two million keystrokes of capacity. That's enough to hold, for comparison, Moby Dick plus the New Testament of the King James Bible).

All of the "Professional" KeyGhost loggers also still have 128-bit encryption of their contents. It wouldn't be very hard for someone who doesn't know the password for a KeyGhost Pro, but who does have some experience with hardware hacking, to dump the entire contents of the Flash memory chip - the actual dump would take almost no time at all, since you're only talking a couple of megabytes for even the top-spec KeyGhosts. But if there isn't some weakness in the encryption scheme, the attacker would then need cubic kilometres of sci-fi nanotech to decrypt the data.

As you'd expect, KeyGhost also now have USB keyloggers for people who prefer a 15-year-old keyboard interface to a 25-year-old one. The USB loggers are more expensive, starting from $US199; the flagship model is $US349. For that price, though, you get a keylogger that date-stamps keyboard activity, and records everything that's typed on any USB keyboard plugged into the computer, whether or not that keyboard's plugged in through the KeyGhost itself. It even works with multiple USB keyboards.

UPDATE: I misunderstood part of the USB keylogger product page. What that part actually meant was that the USB keylogger can be plugged into root ports or into a hub, and still work. It will also work with a keyboard that has its own built-in USB hub, provided all you have plugged into that hub is a mouse (many Mac keyboards are like this). But the USB KeyGhost only logs keystrokes from the one keyboard that's plugged into it.

And then, there's the new "QIDO". It's another little thumb-drive-shaped thing, but it doesn't log keystrokes - it changes them. Its name stands for "Qwerty In, Dvorak Out", and it does what it says on the tin - translates keystrokes from any ordinary Qwerty keyboard into Dvorak Simplified Keyboard keystrokes - and it supports a few different Dvorak variants, too. You activate and deactivate the QIDO by double-tapping Num Lock (or, apparently "Clear", on some Mac keyboards).

If you're one of the few, the proud, the Dvorak-keymap users, you'll be used to fooling around with keymap settings every time you sit down in front of a new computer, and whenever you want to make the computer usable for a Qwerty typist again. With a QIDO, all you need to do is carry the little USB dongle with you. It costs $US119 $US89 plus $US29 delivery, or less if you buy two or more.

The QIDO is a plug-and-play USB device, so to install it, all you have to do is unplug the USB keyboard cable and insert the QIDO between keyboard plug and computer (or USB hub) socket. Actually, because of the QIDO's thumb-drive form factor, I'd recommend you get a little USB extension cable to put between QIDO and computer, so the QIDO isn't hanging in the air, stressing its plug and the computer's socket. But it's still easy to install, and very portable.

The KeyGhost people asked me whether I'd like to review a QIDO, but I don't really see that there's a great deal to review in there. I can tell you now what my review would say: "I plugged the thing inline with a USB keyboard, and the keyboard continued to work normally, except when I tapped Num Lock twice, whereupon I couldn't type any more because I don't know Dvorak."

Ideally, QIDO would magically transform the keyboard's keycaps from "qwerty" to "',.pyf" when you switched modes, but you can only do that if you've got one of those incredibly expensive Optimus Maximus jobbies with a little OLED display built into each key. (The Maximus is apparently quite rubbish to type on, by the way.)

Having the wrong things printed on the keys is not actually a huge problem for Dvorak typists, once they've learned the layout well enough that they don't have to look at the keys for everyday typing, or have just built a mental lookup table of which Qwerty keys correspond to which Dvorak ones.

This isn't as hard as you might think, because standard Dvorak only relocates the alphabetic keys and common punctuation. So the lesser-used symbols of which people are most likely to forget the precise location - @, #, $, % and so on - are still where the keycaps say they are. And if you're learning Dvorak on a Qwerty keyboard you can, of course, just stick a picture of the Dvorak map on the wall and glance at it as necessary.

Since the QIDO can't change the keycaps, though, I was having some trouble figuring out what real advantage it offers over the free alternative - just changing your operating system's keymap.

It's easy to add a Dvorak keymap in Windows - or Mac OS and Linux, for that matter - and then you can switch keymap in a couple of clicks. The QIDO makes switching even faster, but by and large it didn't seem to me that it does anything that changing the keymap in the OS doesn't do.

But then I found this blog post from one Alex Eagle, which I shall now shamelessly plunder.

[KeyGhost now tell me that Alex Eagle is actually "the guy who came up with the concept for the QIDO", so it's obviously not coincidental that his blog-post wish-list so closely matches its actual features.]

Reasons why the QIDO's worth buying:

1: OS keymap control is imperfect. It's possible, for instance, to find certain modifier-key combinations that don't Dvorak-ify properly.

Windows XP (and maybe Vista - I don't know) does Dvorakification in a strange "application-by-application" way. If you add a Dvorak "Keyboard layout/IME" to WinXP, and then bring up the little Language Bar thing and select the new layout, you'll find that you're back in Qwerty mode as soon as you select any other application. This probably isn't what you want, but you're still going to have to separately select Dvorak from the Language Bar for that app, and for every other app you switch to. Each application remembers what keymap is selected, but they all seem to have to be told individually.

Windows Explorer itself counts as an application, here. So you have to select Dvorak after clicking on the desktop or a folder window, if you want to be able to press the-key-usually-known-as-R and have Windows highlight a file whose name starts with the Dvorak-layout P.

I don't think I've quite gotten to the bottom of this, either. The WinXP computer I'm typing this on is now slightly confused, after I switched the keymap back and forth umpteen times; it just switched to Dvorak spontaneously when I was in the middle of typing this document. I can definitely see the attraction of having a keyboard that sends Dvorak-mapped keycodes all by itself, and doesn't even dip a toe into this OS-mediated weirdness.

2: Some software bypasses OS keymap control and looks at direct keyboard scancodes, assuming them to map to the Qwerty values. Or, even more annoyingly, some software may sometimes look at scancodes, and at other times obey OS keyboard remapping. (From reading Raymond Chen's The Old New Thing, I know that just because an application has a user base of more than fifty million people does not mean it won't do boneheaded things like this.)

3: The QIDO lets you have a Dvorak keyboard and a Qwerty keyboard both connected to a computer, and working, at the same time, with no switch-over needed and no fooling with strange WinXP-type keymap selection. This isn't something that most people need, but if you do need it, you probably need it quite badly.

4: Remote computing. If you take control of another computer via VNC or Remote Desktop or whatever, you may or may not get the same keymap at the other end. Again, the QIDO fixes this problem altogether.

You can use the QIDO with any computer you can plug it into, regardless of whether that computer has software support for Dvorak keymaps; it will even work when the computer's not even running a normal operating system, like in BIOS setup programs (provided the computer accepts USB input in BIOS setup, of course) or the Splashtop quick-starting Linux environment. There's probably some allegedly-USB-supporting computer out there that won't work with a QIDO, but it's a standard low-power Human Interface Device, so it really ought to work with just about anything. I could believe it not working if you use it with an old high-power-consumption PS/2 keyboard (like my beloved IBMs), but I wouldn't be surprised if you just needed a better PS/2-to-USB adapter, like the one I mention here.

5: The QIDO doesn't just support Dvorak Standard and a Dvorak-Qwerty hybrid, but also the Single-Handed Left and Single-Handed Right Dvorak variants, for typing using only one hand.

(Certain jokes immediately suggest themselves, but single-handed keyboards of various sorts are immensely helpful for people who only have one hand to type with, because the other one's missing, or because the other one's busy with some other task, like steering their freaky computer-bike, or something.)

You select the keymap you want the QIDO to switch to by using a system taken from the KeyGhosts; type "keydvorak" into a text editor when the QIDO's plugged in, and a "ghost" will type out a menu for you and then await your selection.

Since the QIDO unfortunately does not magically rearrange your keycaps, I think it's likely that most people who'll want a QIDO will also want a keyboard with keys that match their Dvorak layout. It's not easy to actually find an ordinary, inexpensive off-the-shelf keyboard that comes with Dvorak-layout keycaps, but you can often just swap the keycaps around. This'll move the key-locating "pips" that most keyboards have on the F and J keys, and it's unacceptably untidy if your keyboard has differently-angled keys on each rank; if that's the case, you can just use stickers, or break out the sandpaper and permanent marker.

Switching your mind between Dvorak and Qwerty can be a lot harder than switching your keymap. If, for whatever reason, you're better at typing on a Dvorak keyboard than on a Qwerty one - which you'd of course better be at some point in the near future, if you're bothering with Dvorak at all - then you're probably going to need some way of Dvorak-ising any computer you're going to need to type on, lest you overtax your fading brain.

An expensive keyboard with a hardware Qwerty/Dvorak switch on it will solve this problem for you, provided you're happy to carry the darn thing to every computer you use. The QIDO isn't cheap, but it's not as expensive as any switchable keyboard I've found, and it's an awful lot more portable.

The only things it won't Dvorak-ise are computers that can't accept a USB keyboard for whatever reason, and laptops. But you'll probably be able to muddle along with operating-system keymap switching then, if you don't face these situations too often.

I, personally, have not the slightest need for a QIDO. But contrary to my first impression, it really does look like a useful little gadget. If you're using flaky OS keymap switching all the time and tearing your hair out, a QIDO for $US119 plus delivery could be a bargain - and, as mentioned by KeyGhost in the comments below, everybody now gets the $US89 bulk price, even if they're only buying one unit!

A surprisingly un-awful interface

I've become something of a connoisseur of dreadful user interfaces.

They usually grow like pearls, when a programmer adds features to the software he's writing, and sticks the interface elements for each new feature wherever they fit.

From the programmer's point of view everything's fine, because he knows the software back to front and can remember where he's put everything. From anybody else's point of view, though, the interface looks as if a drunk reeled unsteadily around the window, dropping a checkbox here, vomiting up a drop-down menu there. If the program has a lot of features, then even if the programmer doesn't do anything really stupid, the interface can still be hilariously horrible.

It's possible for an interface to be inscrutable at first but pleasant to use after a relatively short learning period - look at Kai Krause's old Photoshop plug-ins, for instance - but developer-made interfaces that just growed like Topsy usually aren't like that.

One that I see quite often is NoteTab's many-tabbed Options dialogue...

NoteTab setup window

...but there are some much more impressive examples.

Bulk Rename Utility interface

Like this, for instance.

I had a directory full of MP3s that all had file names with a number on the front and a chunk on the end that I didn't want, and which all had underscores in place of spaces. I wanted to make the file names prettier.

To do a job like this on a PC, there are four ways you can go.

1: Rename them all by hand. Acceptable for six files, not acceptable for sixty, a tedious way to spend an afternoon for six hundred, a steady job for life for six million.

2: Write a little batch file, usually in some inelegant way like doing a dir >foo.bat and then editing foo.bat so that it ends up as a long list of "ren" commands. Or even uses "for", if you're fancy.

3: Use a proper Unix-style shell like Cygwin or something, that lets you do stuff like this in one operation on the command line, or at worst with a very small script. The standard Unix/Linux/whatever "mv" command can't do this in one line, but any decent shell should have a quite powerful "rename" command. More complex operations are likely to require you to know regular expressions, though.

4: Chicken out and just find a stand-alone file-rename utility.

I chose option 4, and headed off to Pricelessware, the alt.comp.freeware reference archive. Pricelessware pointed me to Bulk Rename Utility, whose inimitable interface you see above.

Let's have another look at it, shall we?

Bulk Rename Utility interface

When I first ran Bulk Rename Utility and saw this, I just sat there and laughed. That lower portion of the window contains, unless I've miscounted, 28 check-boxes, 21 text fields, 17 incrementable numeric-field doohickies each with two increment buttons, 14 drop-down menus, and 17 other miscellaneous buttons. And it's got a normal complement of ordinary menus up at the top!

The hell of it is, though, that this interface is actually very usable. It works, and it works well. I learned how to make it do what I wanted in, I don't know, maybe 30 seconds. It could probably be better, but it's by no means actually bad.

Bulk Rename Utility would be every bit the nightmare it appears to be, if it weren't for one very sensible move on the programmer's part: The interface shows you, in green, a preview of what your instructions are going to do to whatever files you've selected.

So I could instantly see that my search-for-"_"-and-replace-with-" " operation was going to annihilate all of the filenames altogether, and I said "huh?", and then I noticed that I'd accidentally put the search-and-replace terms in the "RegEx (1)" part of the interface instead of the "Repl. (3)" part, and I fixed that, and it was fixed.

Capitalising words in filenames is easy, too - just stare at the interface for 10 seconds to find the right bit of it ("Case (4)"), and select the option you want - "Lower", "Upper", "Title" or "Sentence" - from the menu. If those names don't immediately explain their function to you, the green preview will.

Bulk Rename Utility even deals elegantly with locked files. If you choose to abort the renaming procedure because a locked file can't be renamed (in this case, the file was still open in my music-player), Bulk Rename Utility gives you the option of reversing all of the renames it did before getting to the locked file.

Using regular expressions via this interface isn't much easier than using them on the command line, but there are umpteen other options for everyday, and some quite unusual, rename operations. If you're not doing something pretty darn complicated, you ought to be able to get it done with the standard interface.

An interface like this is still a usability disaster for a lot of people, though.

Pretty much anybody should be able to learn how to do complex rename operations with Bulk Rename Utility if they just spend a couple of minutes playing with it. OK, you might have to look at the help, or even download the manual in PDF format. But it's really not that difficult, even if you're not good with computers.

But your average computer-phobic person won't even try to use something that looks like this. The same explosion-in-an-interface-factory quality that cracked me up when I first saw Bulk Rename Utility has a much more negative effect on people who aren't confident about using computers.

Sometimes that lack of confidence is justified. It's still easy to find big-name software products, including whole operating systems, that just don't bloody work for some tasks, or that drop dead unexpectedly and then need a lot of work to fix. One of the standard things you hear from computer-phobic people is "I don't want to mess around with it, in case I break something". If that fear is based on a memory of a time when something really did break, then it'll be a difficult phobia to cure.

But many computer-phobics haven't had many, or any, experiences of this sort. They just don't know much about computers, and have decided that this means they will never know much about them. I find that attitude very frustrating - "give a man a fish and he'll eat for a day, try to teach a man to fish and he'll tell you you're wasting your time".

On the plus side, Bulk Rename Utility might actually serve as a sort of exposure therapy for computer phobias. Show it to the computer-hater, let them soak up how completely impossible it looks, then let them use it to rename a directory full of temp files. Whaddaya know - it's not that hard, if you just read the little labels and then see what they do!

The only thing wrong with Bulk Rename Utility for this purpose is that renaming multiple files isn't a task that ordinary users actually have to do very often. Ideally, you'd want a daunting-seeming, actually-easy-to-use program that greatly speeds up some painful task that ordinary users do often have to do.

Any suggestions?

Are LED flashlight years like dog years?

Elderly Arc-AAA

This is the Arc-AAA LED flashlight I reviewed back in 2001. It's been in my pocket pretty much constantly since then, which is why it's now more silver than black. I suppose it now qualifies as a "vintage" LED flashlight.

(I've still got the Arc-AAA Limited Edition somewhere, too. I hardly used that one at all - it might be worth something to some nut enthusiast on eBay. The Arc-LS still works A-OK, too, though its rubber switch-boot perished some time ago. It's happy as a clam with a rechargeable CR123 cells, just like the Mr Bulk LionCub that came out when "RCR123s" were still a bit exotic.)

There's nothing to break in an Arc-AAA. The only maintenance it needs is an occasional dab of fresh solder on the contact on the back of the lamp assembly. Its on/off "switch" is of the simple turn-the-lamp-head type; you might think that the screw threads might wear out, on a flashlight like this one that's made of aluminium. But it still seems fine to me. (I've cleaned and oiled the threads pretty often, to keep abrasive crud off the threads; an unmaintained Arc-AAA would probably be pretty dodgy by now.)

The LED itself should just very slowly dim as its hours of use mount up. Extremely slowly, actually, for an LED like this that isn't on for more than a few minutes a day, and isn't even driven particularly hard except when the battery's brand new.

So my Arc may still work when it's as old as my genuinely elderly flashlight.

In the years since I reviewed the Arc-AAA, Arc Flashlight went broke and were, a while later, reborn under new management. They now sell an updated Arc-AAA, plus an excitingly expensive light called the Arc6.

The current Arc-AAA is probably quite a lot brighter than my old one, with the same or better battery life. White-LED lumens-per-watt have improved very fast over the last ten years. I bet you don't even need a soldering iron to keep the lamp contact shiny any more.

You no longer have to buy an Arc or a Peak if you want a single-AAA-cell LED flashlight, either. There are umpteen other on- and off-brand options in the 1xAAA size. A lot of them have a "one watt" super-LED instead of a 0.1-watt-ish 5mm unit, too. A one-watter in a 1xAAA light will probably be running at a fraction of its rated power, or else it'll frighten the battery to death in no time. But single-AAA "one watt" lights will probably still give you a lot more light than even a modern 5mm LED is likely to manage.

But I still like my old Arc.

Ultracapacitors versus batteries, no holds barred!

A reader asks:

Do you have any thoughts on the "Light for Life" capacitor-powered flashlight?

Light For Life

Neat development, or doomed to failure?


In brief: Yes, this flashlight probably works well enough (or will probably work well enough - it's not quite on sale yet).

There are some big differences between this "Light for Life" and conventional battery-powered LED flashlights, though, all because of the LfL's unique selling point, its electric double-layer capacitor, a.k.a. "ultracapacitor", power source.

This post's going to go on and on and on, which is a bit ridiculous for a "review" of a product that I've never actually even seen in the flesh. But I found crunching the numbers for ultracaps - also known as "supercapacitors" - versus conventional batteries quite interesting. Ultracaps are just starting to break through into the realm of actual consumer products - high-charge-rate regenerative braking reservoirs for electric cars, for instance - so I took this as an opportunity to see just how close they've come to regular-battery capabilities.

First, though: There are two basic ways in which a new flashlight product can be a scam.

(Well, there are two ways if you don't count "there is no actual product, we just take customers' money and never send them anything", or some similar blatant fraud.)

The first kind of scam is to use misleading specifications. In the case of a flashlight like this one that runs from capacitors and is thus likely to have much less run-time than a similarly bright flashlight with batteries in it (exactly how much less, I'll address shortly), you could for instance make it sound better by focusing its output into a really tight spotlight beam. Then you'd specify the light's brightness in candelas (or millicandelas if you want a really big number), without mentioning that the tighter the beam, the more "candelas" you get per unit of actual light output.

The second popular scam technique is what I like to call the Electric Car Trick. Have you noticed that every news piece about an electric car says something like "it can do 150 miles an hour, and has a range of 200 miles"?

A more accurate statement would be "it can do 150 miles an hour, OR go 200 miles, if you drive it slower". Conventional engines can't simultaneously deliver maximum power and maximum fuel economy, and neither can electric cars. (The fuel tank in a Bugatti Veyron, to pick a fun example, holds a hefty hundred litres of petrol, giving the car a range of more than 400 kilometres if you drive it gently. But the fuel tank will be empty after only twelve minutes of full-throttle driving, so even though the Veyron can do about half the speed of light, it's got a full-throttle range of no more than 75 kilometres.)

The flashlight version of this trick involves giving the light a low-brightness mode and a high-brightness one, and quoting the light's brightness in the high-output mode, and its run time in the low-output mode, on the sales brochure.

Or, if you've got a simple "passive" LED flashlight that just hooks the batteries straight up to the lamp (usually through a current-limiting resistor), you can specify the light's brightness as whatever it delivers in the first five seconds of the life of a brand new set of batteries, and the run time as many, many hours, for almost all of which the flashlight will not actually be anything like that bright.

A simple batteries-and-resistor LED flashlight like this will give better-than-nothing light for ages and ages on almost-dead batteries. If the manufacturer's audacious, they can decide that "run time" lasts until the light's output is 1% of what it was to start with. Consumers are unlikely to agree with this decision.

(The Light for Life pretty much has to be a "smart" flashlight with a current-and-voltage-controlling driver circuit, because if you plug an LED straight into a capacitor through a current-limiting resistor, the LED's brightness will fall as the capacitor empties. Batteries have pretty steady output voltage as they flatten, but the terminal voltage of a capacitor is directly related to its state of charge.)

These sorts of shenanigans can be spotted pretty easily by seeing if the spec-sheet numbers add up, and in the Light for Life's case, I'm happy to say that they do. I don't know the specifications for the ultracapacitors in the LfL, and I also don't know how much power's lost in the circuitry between the caps and the LEDs, but I can still do some back-of-an-envelope calculations.

First, I looked up a random off-the-shelf ultracapacitor, to see what kind of performance they're offering. For about $US35, you can get a cylindrical ultracap 61 by 81 millimetres in size (a D battery is about 34 by 60mm), weighing 405 grams (D battery: about 150 grams), and with a capacitance of a whopping 1200 farads at 2.7 volts.

Because capacitors only deliver their full rated voltage when they're completely charged, you can't just multiply the capacitance by the voltage to get the nominal energy content of the fully-charged cap. Instead, you use this formula (where C is capacitance in farads, V is voltage in volts and E is energy in joules):

0.5 times C times V-squared equals E.

So if you've got an ordinary sort of cigar-butt-sized electrolytic capacitor - not a super- or ultra-capacitor - with a rating of 680 microfarads at 35 volts, when it's fully charged its capacity is

0.5 * 0.00068 * 35^2

...which gives energy of 0.4165 joules. A joule is a watt-second, so this capacitor could deliver one watt for 0.4165 seconds, or 0.4165 watts for one second, or any other combination in which watts times seconds equals 0.4165.

The other day, I was digging through some junk and found my monstrous old electrolytic capacitor, the size of a beer can - it's about 13 by 7 centimetres, not including the screw terminals on the top.

I've no idea how much of its original capacity this paperweight-cap has retained, but it's rated at 850 microfarads at 450 volts, which are very impressive numbers for a non-super-cap.

So for this cap, the calculation goes

0.5 * 0.00085 * 450^2

...which equals 86.0625 joules.

For comparison, if someone with a very good arm throws a cricket ball at you at a hundred miles an hour, that ball leaves the thrower's hand with a kinetic energy of about 160 joules.

(The calculation for a baseball would be much the same; cricket balls are a little heavier and rather harder than baseballs, but fast-bowlers, on average, bowl a bit slower than fast-pitchers pitch.)

If you'd prefer the Guns and Ammo comparison, 86 joules is about the muzzle energy of the weedy .25 ACP pistol cartridge.

Right. On to the that bigger-than-a-D-cell ultracap, with its monster 1200-farad capacitance at a meagre 2.7 volts:

0.5 * 1200 * 2.7^2 = 4374 joules.

That is a LOT, by the standards of everyday non-"super"-capacitors that human beings can lift. And the ultracap can deliver it pretty quickly; the spec sheet says its maximum discharge power is more than 3100 watts.

Ordinary, non-"super", capacitors can be charged and discharged in a tiny fraction of a second without harming them, but ultracapacitors can't. If you just drop a screwdriver across the terminals of this ultracap it may discharge very fast indeed, and give you a frightening lesson in what 4000-odd joules in a tiny fraction of a second means (a .303 cartridge has a muzzle energy of only about 2500 joules). But don't expect the ultracap to be healthy after the smoke clears.

So how does this 4374-joule capacitor compare with a battery, say an ordinary low-self-discharge NiMH AA with 2.2 amp-hour capacity, for flashlight-powering purposes?

Not well.

The NiMH cell's nominal 1.2-volt output (which it will actually deliver for almost all of its discharge cycle, unless it's very heavily loaded), times 2.2 amp-hours, gives 2.64 watt-hours. There are 3600 seconds in an hour, so 2.64 watt-hours is 9504 watt-seconds, and a watt-second is a joule. So this one unremarkable AA cell has more than twice as much energy storage capacity as the much larger ultracapacitor.

A NiMH AA will only weigh around 30 grams. So for the 405-gram weight of the ultracap, you could have about thirteen AAs.

But you definitely can't charge nickel metal hydride batteries - or any other rechargeable battery, for that matter - in ninety seconds, which is one of the Light for Life's big selling points. If you want your NiMH cells to live a long and happy life, one hour is about the fastest charge they can take. (There are "15 minute charge" batteries out there, but they haven't really taken off, on account of how they cheat a bit.)

But once a conventional rechargeable is charged, it can run a modest load, like a flashlight, for much longer than any capacitor yet made.

(Now it's not so surprising that there's an accessory for the Light for Life that lets you run it from AAs instead of its capacitors.)

So: How does the Light for Life, running from its standard ultracapacitor "battery", stack up against conventional LED flashlights?

The LfL manufacturers say (there's a bit more detail in the PDF brochure, here) that the Light for Life has a 90-lumen "standard" mode, and a 270-lumen maximum-power mode (plus a "tactical strobe" mode that flashes the maximum-brightness beam, to disorient an attacker). It's also got a 25-lumen "standby mode", which is all you get when the light's running out of juice.

The lumen is not a scam-friendly unit; a lamp with 270 lumens of output has the same 270 lumens no matter how tight the beam is or how close you put the light-meter. The Light for Life does have a rather tight beam, though; they say the main beam covers a spot about 22 inches in diameter at a range of 20 feet, which adds up to a main-beam width of only about 5.3 degrees.

I've seen narrower-beamed lights than this - the tiny Weiguo Solutions Spotlight, for instance, has a main beam width of only about four degrees. But the LfL beam is still narrower than you want for an everyday seeing-where-you're-going sort of flashlight. But a narrow beam is, of course, superior if you want to see things at a distance.

The Light for Life apparently does have about a 15-foot "corona" around the main beam at 20 feet, which is about a 41-degree width and means the light will be perfectly usable for everyday non-possum-spotting kinds of tasks. But if you want to see everything in the room at once, it's not great.

In full-brightness mode, the Light for Life is apparently good for a run time of 15 minutes, after which it'll drop to the 25-lumen mode, which it'll apparently be able to sustain for another 30 minutes. The "standard" 90-lumen mode is said to be good for 60 minutes, plus the same 30 minutes of 25 lumens.

I don't know what the actual capacity of the ultracapacitors inside the Light for Life is, but I can make a guess based on these figures.

Good commercially-available white LEDs currently have a luminous efficacy of about a hundred lumens per watt.

15 minutes of 270 lumens at 100 lumens per watt is 40.5 watt-minutes, which is 2430 watt-seconds, or joules, of output light energy.

60 minutes of 90 lumens at 100 lumens per watt is 54 watt-minutes, which is 3240 joules.

These numbers aren't the same because, I bet, the 90-lumen mode is running the LEDs at moderate power, when they give the best efficiency; in the 270-lumen mode the LEDs are being pushed harder, so their efficiency falls and you don't actually get 100 lumens per watt - or whatever their actual rated luminous efficacy is - out of them any more.

So let's use the 90-lumen "standard" mode for energy-measurement purposes, since that's where the Light for Life is probably at its best.

The LEDs should be just as efficient, if not more so, in the low-powered 25-lumen "standby" mode; it's just not bright enough to be very interesting by itself. LEDs are generally perfectly happy to be driven at a lot less than their rated power, which is why simple LED flashlights have such immensely long better-than-nothing light output from almost-dead batteries.

But let's assume, for simplicity, that in the standby mode the luminous efficacy stays at 100 lumens per watt. In that case, 30 minutes of 25-lumen light at 100 lumens per watt gives 7.5 watt-minutes, or another 450 joules.

So ignoring other losses in the system, the total energy you can wring out of the ultracaps with a full-duration standard-brightness run followed by a full-duration standby-brightness run is 3240 plus 450 joules, for a total of 3690 joules.

These numbers look fine to me. Since this flashlight is about the size of one that runs on a few D cells, it'd be no problem at all to pack that much ultracapacitor energy storage into it. I wouldn't be surprised if the total energy capacity was about 4500 joules. Maybe even more, if the driver hardware isn't terribly efficient.

But, as mentioned above, a single humble NiMH AA cell will give you over nine thousand joules. It could even be drained in 15 minutes without damage, so even after losing some efficiency by plugging your AA into a voltage-booster to allow it to run white LEDs (which want about 3.6 volts), that one lousy AA cell could quite easily give you twice the run time of the Light for Life ultracapacitors.

Upgrade to a few NiMH D cells, which you can easily fit in a flashlight the size of the Light for Life, and the comparison becomes ridiculous. It's easy to find NiMH D cells with a capacity of ten amp-hours; three of those would give you about 35 times the energy storage of the Light for Life's capacitors.

There are lots of LED flashlights with light output up there with the LfL in its maximum brightness mode. Boutique manufacturer Elektro Lumens, for instance, currently offers a retrofitted 3-D-cell Mag-Lite with a four-die LED in it, that blasts out "up to 930 lumens" from rechargeable-D-cell power, for $US129.99.

Picking another LED flashlight manufacturer more or less at random, Peak LED Solutions will sell you a durable little "over 220 lumen" flashlight called the "Night Patrol" for $US95, plus the price of the single 18650-sized lithium-ion battery and charger it runs from (those batteries and chargers are very cheap these days, as I mentioned in this review).

And then, of course, there's Mag Instrument, the makers of the iconic Mag-Lite. They took forever to start making LED flashlights - umpteen people, like Elektro Lumens, made drop-in LED lamps for Mag products in the meantime - but now they finally do. Their LED flashlights (and LED upgrade kits for some other Mag flashlights) all use a "three watt" LED of not-especially-cutting-edge quality, so they don't compare very well with other LED flashlights. You'd think they'd have an output of 200-odd lumens, maybe 250 from fresh batteries, but even the big D-cell Mag-LEDs don't seem likely to beat the Light for Life's "standard" brightness, ever.

The Mag options are solidly made and cheap, though, and just about any batteries - even cheap "super heavy duty" carbon-zinc cells - will give them far better run time.

The 3-D-cell "Mag-LED" flashlight is in the same size class of the Light for Life, and should have a run time of an easy nine hours from three 10Ah rechargeables. The Light for Life should weigh only about half as much as the Mag-LED with batteries, though. That's good if you're carrying a bunch of gear already, but not so good if you think you may need to bludgeon someone with your flashlight.

The list price of the Light for Life, as per the PDF brochure, is $US169.99. It comes with a charger and the capacitors, of course, so it's unfair to compare it with something like a Mag-LED that doesn't necessarily even come with non-rechargeable batteries, much less rechargeable ones.

To have something like the same 90-second charge-and-go convenience as the LfL, you'd actually need two sets of batteries for a conventional rechargeable flashlight. And a charger, of course.

So. 3-D-cell Mag-LED: List price $US32.99, yours for about $US21 ex delivery from various dealers.

Six quality NiMH D rechargeables: About ninety US dollars, for three 2-packs of name-brand cells, or about fifty bucks for six 7Ah no-brand cells.

There are also suspiciously cheap D cells that're actually just smaller cells in a D-sized wrapper; if you buy those, you could get away for $30 or so and still have way more joules than the Light for Life offers. You can also buy "spacers" that you put a AA cell in to make it the size of a larger battery; six quality AAs plus six spacers will cost you only about $US27. But "proper" D rechargeables cost less per joule than any of these options.

And, finally, the conventional-flashlight option also requires a quality D-cell-capable charger like a Maha C808M (which I recommend Australian buyers purchase from m'verygoodfriend Jeff Servaas). That's about $US93 delivered for American shoppers ($AU183.15 delivered from Jeff; that's about $US125, as I write this).

Total US price for all this stuff: Probably a bit more than $US200, including delivery.

In the final analysis, it's simple enough to figure out whether the Light for Life is for you. It genuinely does seem to be as bright as other flashlights in its size and (total) price class; about as bright as the spiffy tweaky ones if only 15 minutes of run time doesn't bother you, and about as bright as a cheap Mag-LED if you'd prefer 60 minutes before you enter "standby" limp-home 25-lumen mode. And an hour of run time at Mag-LED brightness is more than enough for most purposes, and the LfL probably really does recharge in no time at all, and it probably really will last quite a long time if you don't horribly abuse it, because ultracapacitors don't wear out like batteries.

It's always possible that the LfL people are overstating their product's abilities or have made some other terrible mistake, like the Altus Lumen people did. But if the spec sheet's telling the truth, this looks like a good product for a reasonable price to me.

(Now, of course, to scam one for review. I'm not sure whether the fact that I've pretty much already written the review will count for or against me.)

Zwei Keyboarden

I've accidentally become some sort of Internet authority on clicky IBM-type keyboards. So I've been receiving a pitter-patter of e-mails about the latest products from Das Keyboard.

Das Keyboard originally sprang to nerd-fame with their first, eponymous model, which was notable for what it lacked rather than what it had:

There was nothing printed on the keys.

Personally, I think that's goofy. I don't have to look at the keys when I'm typing, but that doesn't mean I can always remember which key has % and which has ^, especially after the second martini.

I suppose an un-marked keyboard could be of some value as a training aid, and it certainly does have a unique aesthetic appeal. But if you haven't memorised all of the keys, including the used-once-in-a-blue-moon symbols, then to avoid having to just guess when you need to type something unusual on your blank 'board, you'll have to hang a picture of a normal keyboard on the wall.

(I can, by the way, type something in the order of 80 words per minute, which makes a very impressive 900-odd clicks per minute when I'm using a buckling-spring keyboard that makes two clicks per keystroke. I brought my own keyboard to work back when I worked for the Dark Lord Murdoch, and enjoyed an entirely unjustified reputation for doing exactly what I was supposed to do.)

The original Das Keyboard didn't have much to commend it besides its slick appearance, because it was a rebadged Keytronic membrane 'board. It was about as good-feeling as a membrane keyboard can be, but it cost well over twice as much as the printed keyboard it was based on.

Then they made the Das Keyboard II, which was a proper clicky keyboard with discrete keyswitches. Except they didn't really "make" it; I think the II was another rebadge job, this time based on the Ione Scorpius M10. Which, once again, was cheaper than the Das Keyboard version.

Time marches on, and Das Keyboard now have two mechanically-identical keyboards. The first is the "Ultimate"...

Das Keyboard Ultimate

...which is another blank 'board, and the second is called the "Professional"...

Das Keyboard Professional

...which - gasp! - has normal printed keys.

(This reminds me of the Penguin caffeinated-mint company, and their "decaffeinated" mints.)

I'd be happy to review a Das Keyboard Professional, but the last time I dealt with Das Keyboard they were apparently playing she-loves-me, she-loves-me-not, in a variant called we'll-tell-Dan-we're-sending-him-a-keyboard-for-review, no-we-won't, yes-we-will...

This lengthy process ended on "no-we-won't".

I don't think they're actually trying to hide anything - reviews of the new models have been very positive. The only real question is whether, aesthetics aside, you can get something just as good for less money.

It seems that the Ultimate and Professional actually are the first Das Keyboards that you can't buy under another name. If they are still rebadge jobs, I can't find the original models this time. The new 'boards have Cherry keyswitches, so I suspect they're being made by Cherry. But nothing in the Cherry keyboard lineup looks like the new Das Keyboards; there's a "smart card keyboard" that looks a little like them (it has the same projection in the top right corner, which is where the smart-card reader lives), but the key layout is different.

The new Das Keyboards aren't cheap. They list for $US130 in the States, but they're about as big and heavy as an IBM buckling-spring keyboard, so you probably don't want to buy them from overseas. Here in Australia, you can get them from Aus PC Market for $AU198 delivered to anywhere in the country; Australian shoppers who'd like to order the unprinted Ultimate can click here do so, while the printed Professional is here.

The current exchange rate actually makes the keyboards a little cheaper than the US price; as I write this, 198 Australian dollars is only about 122 US dollars, and the Aussie price includes delivery. And any clicky keyboard is likely to last a long time, so a couple of hundred bucks isn't really that much to spend.

(Note that currency exchange rates are unusually variable at the moment, thanks to the financiapocalypse currently sweeping the world. If you're reading this only a month or two after I wrote it, don't be surprised if exchange rates are vastly different.)

There are indeed, however, other clicky-keyboard options.

A few years ago, almost nobody was making clicky keyboards any more, but there's been a resurgence lately. Keyboard connoisseurs are used to fossicking through new-old-stock dealers, used 'boards on eBay from sellers of variable honesty, and of course Unicomp. But there are now a few other companies making keyswitch keyboards.

The most impressive "reborn" keyboards on the market today, if you ask me, are CVT's Avant Prime and Avant Stellar. They sell for $US149 and $US189 respectively, but that's because you can remap almost every single key, and also bind macros to arbitrary keys. Actually doing this is less than totally straightforward, because the CVT 'boards are clones of the old Northgate OmniKeys, and work the same way.

There's also Deck, who made the little keyboard I reviewed a while ago. They have a full-sized IBM-layout 'board called the "Legend", but it's $US149 ex shipping. (The little "Deck 82", like the one I reviewed, is $US99.) All of the Deck 'boards have LED backlighting, though, which really is quite fun.

(On the subject of mini-keyboards: If you're looking for a small decent-feeling non-clicky membrane keyboard, you could also check out the "Happy Hacking Keyboard", which has been available in several different models, even including a blank-keytop version. The only one available now is the $US69 "Lite 2", though.)

There's also Ione (or iOne, or whatever they want to be called), the makers of the Scorpius keyboard that was rebadged as the Das Keyboard II. They're still making the Scorpius M10 It has no fancy features at all, but can be had for fifty US bucks, which is hard to beat.

As I write this, there are three Amazon reviewers complaining about Scorpius keyboards with lousy build quality, but other reviewers specifically mentioned how well the keyboard was made, so I'm not sure what's going on there. Nobody seems to sell the M10 here in Australia, so I don't anticipate getting one to play with any time soon.

SteelSeries have two non-clicky discrete-keyswitch keyboards. That's what you want if you're after good tactile feedback - which lets you type faster and with less effort - but don't want a 'board that makes a racket. Their SteelSeries 6G lists for 99.99 Euros (about $US125, as I write this) plus shipping, and their SteelSeries 7G is 129.99 Euros. The main difference between the two is that the 7G has audio connectors and controls and apparently lets you press every key at once without any being lost, while the 6G has a more normal eight-keys-at-once buffer, and comes with a bunch of grey keytops that you can swap in to make important keys stand out.

There's also the Gigabyte GK-K8000, which is a bit unsightly but has Cherry keyswitches, a bunch of extra programmable keys, and onboard USB audio. It apparently lists for $US113, but doesn't actually seem to have quite made it to the shelves yet.

Mac users might like to check out the Matias Tactile Pro, which resurrects the old Apple Alps-keyswitch keyboards. The Tactile Pro version 2 sells for $US149.95, plus at least $US20 shipping.

If I were shopping for a clicky keyboard right now, my first stop would be eBay, to see if someone within inexpensive-shipping-distance of me had a decent-looking buckling-spring IBM 'board, or maybe something with the classic Alps keyswitches. If you're in the USA, you're very likely to be able to find a buckling-spring or discrete-keyswitch keyboard with many years of service left in it for less than fifty bucks including delivery. and Unicomp are excellent options for US shoppers, too.

If you're somewhere like Australia, though, you'll probably be waiting a while for your saved eBay search to turn up any options, and shipping prices for battleship keyboards from overseas will be painful.

Now that keyswitch keyboards are normal retail items again, you might as well just get yourself something like the Das Keyboard Professional. It won't cost you any more, you'll get a local warranty, and you'll probably be delighted.

You need a Lego earthmover

Lego 8294

I had the Lego #8851 Pneumatic Excavator when I was a kid (and still have all the parts, natch), so when I noticed that Kmart here in Australia is currently selling the new and exciting #8294 linear-actuator Excavator for only $AU54 (US list price $US60!), I had to get one.

(OK, actually I got more than one. They also have the bigger #8295 Telescopic Handler for only $AU89 - list price $US90 in the States. The Handler only has one linear actuator in it, though; the Excavator has two. The sale's on until the 8th of October.)

Lego 8294 reaches out

Because the Excavator has only two actuators - the old pneumatic one had three - its bucket-hinge action is linked to the end segment of the arm. This makes it a bit less playable.

(The linear actuators are part of the new "Power Functions" motorised-model line, but are not themselves motorised unless you buy extra stuff. They're discussed in great detail on the excellent Technic Bricks.)

Apart from that, though, the new excavator is brilliant. I miss the more expensive old-style packaging you used to get with Lego; now each set is just a flimsy box full of bags. But I don't miss the old studded-beams Technic Lego itself at all. The new stuff makes it much easier to pack tons of mechanism into a small space, and if you just chug through the instructions without spending a lot of time puzzling over what in fact it is that you're building, it's a wonderful surprise when you stick it all together and suddenly find yourself looking at a freakin' gearbox.

Lego 8294 gearbox

The gearbox uses some specialised parts that've been around for years, but were new to me. In this case, they give you a shift lever with a neutral position in the middle, and either end of its throw linking a gear on the back of the Excavator to one of the two actuators.

(The actuators are powerful and accurate, but not what you'd call speedy. Frequent users may like to replace the gear on the back with a less pretty but more usable crank, or an electric motor - the instructions have a bit at the end that shows you how to add a Power Functions motor to the set. This, by the way, explains the funny little peg sticking up on one side of the tracked base; it restricts rotation of the top of the excavator for no purpose in the standard model, but if you add the motor, the turn-stop prevents you from twisting up the wire going from the battery pack in the base to the motor in the top part.)

New chunky Lego tracks

The new excavator has those new chunky tracks I was talking about the other day. Unlike the somewhat fragile old-style gear-drive tracks, the new ones are deliberately made to not hook together terribly strongly. So if you twist the track a bit, one of the links will click apart. That may put a kink in your plan to use these tracks for heavy-duty off-road motoring, but it may also cause your new Lego Panzer V to throw tracks about as easily as the real one did.

The #8294 excavator also comes with several stickers you're supposed to put on the pieces; I of course did not even glance at them. And it's got this unusual giant tile piece, which is only used in the alternate model. Lego have cut costs here, as well; you don't get instructions for the second model in the box, but have to download them instead.

I'd normally complain about this, since downloadable extras can reasonably be expected to not be available ten years down the track. But there's pretty much zero chance that it'll happen in this case; the fan community will provide, if Lego ever don't.