Mindboggling

Last week's weather was really ugly at times, and in particular, we had one night when it rained, then snowed, then rained, then snowed ...  It ended up with the driveway covered several inches deep in slush too heavy for the snowblower to shift.  Then the slush froze, leaving an ice crust thick enough to drive on.  I wound up, in desperation, hooking up the hosepipe to a hot-water faucet to get the ice off the driveway.  (This wasn't as absurdly energy-intensive as it sounds.  The ice was spongy, and melted very rapidly under a strong jet of hot water.

While I was doing this, one of our neighbors said, "You know hot water freezes faster than cold water, right?  Try it in your freezer sometime, put an ice tray of hot water and one of cold in there, and the hot water will freeze first."

I was so gobsmacked I just didn't know what to say.  I've heard some ridiculous urban legends in the past, but this one ... I just don't know where to start.  Where on earth does anyone get such a totally wrong-headed idea?  It falls down on even the very simplest of examination — the hot water's got to get cold before it can freeze, and that's going to take time, so the only way the hot water can freeze faster is if some cold water freezes faster than other water, but how do you know you didn't get magic fast-freezing water out of the cold tap in the first place?  To freeze before the water that started out cold, with both trays being cooled at the same rate, the hot water would have to cool to the starting temperature of the cold water in negative time.

Snopes is silent on this one.

The only way I can think of that someone could "prove"this to themself is if they inadvertently put the tray of hot water right under the cold-air inlet to the freezer compartment, so that it's being much more strongly cooled than the other.

Pondering patents

For no particular reason that I can discern, I found myself thinking about patents and patent law this morning, and thinking about what's wrong with the current system and how to fix it.  I came up with seven basic points that a "good" patent law should cover:

1. Patents must be specific

   You can't file a patent based on vague generalizations and handwavium.  You have to be specific about what you're patenting.  You can, for example, patent a clever new way of laying out the gearsets and shafts in an automatic transmission that lets you get seven forward speeds from the sane number of planetary gearsets you're currently using to get five, or a new tread compound or tire belt construction that gives a tire lower rolling resistance for the same traction and tread life.  But you can't patent the idea of a low-rolling-resistance tire or a seven-speed automatic transmission.

2. Patents must be non-obvious

   Patents are intended to reward innovation ... so innovate.  You can patent a new design of saw tooth that cuts faster with less effort, or a new process for heat-treating a saw blade so that it'll cut longer without dulling or shipping, but it's a bit late to patent the saw.  If someone else invents a clever machine that lets a truck carrying the equipment quickly and efficiently de-ice roads, you can't patent a bigger version of the same machine that de-ices two lanes at a time (or a smaller version that de-ices sidewalks), and you can't patent the idea of using the machine to de-ice airport runways.  (However, you MAY be able to patent an ingenious way of making the mechanism smaller that enables building one small enough to clear sidewalks.)  Similarly, you can't patent something that's just a trivial combination of existing patents.

3. Patents must be tangible

   You have to be able to point to a specific thing you're patenting.  It can be a device, an algorithm, a manufacturing process, a specific use of something, but you can't patent abstract ideas such as business methods ... you can't patent, say, having separate accounts-due and accounts-receivable departments, or keeping your customer information in a CRM database, or building sedans and station wagons on the same line.

4. No patenting Nature's inventions

   You can't patent a tree, a bacterium you found, or a gene you sequenced from someone's DNA.  However, if you developed a new drug from a compound you found in the tree, or you developed a novel medical treatment based on what you learned from studying how that gene operates, you can patent that.

5. No submarine patents

   You can file a preliminary patent on something you haven't worked out all the details of, but you get to file three revisions of your preliminary patent in two years, and if you don't have something solid enough to file a real patent on by then, it's thrown out and you have to start over.  You can't keep on amending and amending the same preliminary patent forever, under the radar, until someone stumbles into it.

6. Use it or lose it

   You can't just sit on a patent and wait for people to infringe it, then litigate.  If you are granted a patent, you must either exercise it, or license it.  If you won't use it yourself, and won't license it, but just sit on it to prevent other people from using it or so that you can sue people who unknowingly infringe it, you lose the patent.  The purpose of patents is to spur innovation, not to suppress it.

   (Addendum:  No-one should ever be penalized for declining to license an invention to an unfriendly foreign power or agents thereof.)

7. Enforcement and disclosure

   It's your responsibility to enforce your patent.  But if you discover that someone is inadvertently infringing your patent, you may not litigate without first making a good-faith effort to license the patent to them.  The other side of the coin is, you may not knowingly withhold discolsure of a patent — if you are engaged in efforts to license a patent that you know imples or requires another patent held by you or a business partner, or if you are trying to submit a patented technology to a standards body (to cite two examples), you must disclose all applicable patents.  If you fail to disclose a patent in such a situation, you must either grant a free license to everyone involved, or you lose the patent.

Thoughts?  Comments?  Did I miss anything major?

About that Linux kernel vulnerability

Most folks who care have probably heard about this Linux kernel vulnerability by now.  And a lot of people out there are probably thinking, "Ah, but it's a local exploit, I'm safe as long as I don't give access to anyone I don't trust."  Even securitywire says, "It may not be remotely exploitable, but [...]"

But remember, what this means is that now, a remote attacker doesn't have to get a privileged shell.  All they have to do is get a shell ... any shell ... with the ability to (via one route or another) place a compiled copy of the exploit code in, say, /tmp and execute it.  To quote Mari Nichols of SANS,

"I believe Secunia has correctly identified this vulnerability as a local system vulnerability, but given that every server with a vulnerable kernel can be exploited to get elevated privilege, any unprivileged remote exploit can combine with it to form a remote root-level exploit."

Making a difference

On Babylon 5, on his way to apologize to the Narn people for the wrongs done them by the Centauri, the Centauri Emperor Turhan said,

"The past tempts us, the present confuses us, and the future frightens us... and our lives slip away, moment by moment, lost in that vast terrible in-between.  But there is still time to seize that one last fragile moment, to choose something better, to make a difference."

Australia has decided it is time to make a difference.