For the last 15 years and more, I have been dealing with crashes.
Not plane, car, or train crashes, but computer crashes. For many of those years, it was as the Q&A guy for MacUser. For some of them, it was also trying to support my own apps in the amazingly crash-prone hands of sailmakers.
The OED considers that crash is an onomatopœa, enjoying the same relation with crack that clash has to clack and clap, an impressive piece of word-play for a dictionary. It also points out that modern Scandinavian languages have analogous words, such as the Danish krase, to crackle.
Its literal uses generally invoke sounds similar to that of the word itself, although in the early nineteenth century it acquired a figurative sense of falling into ruin suddenly and violently, particularly in a financial way. It does not appear to have been used of motor vehicle (or aircraft) accidents until the early twentieth century, and its first use in computing is recorded as being in 1972.
In these days of primetime TV A&E/ER soaps, we now seem to prefer the acronyms of the emergency services to describe vehicle crashes, and crash is mostly associated with aircraft impacting into the ground. Oddly it has rarely been used in maritime incidents, except by reporters who are unfamiliar with nautical parlance, although you will sometimes hear of vessels crashing into immovable objects.
I wondered whether its use in computing originated in the physical and audible crashes which used to be common in hard disks. As they were first invented at IBM in 1954, that is plausible, but the classic Winchester drive which certainly did crash on occasion did not appear until 1972, when the word was already in use to describe more generic computer failure.
Not too long ago, hard disks did onomatopœically crash, when their read/write heads made contact with the platters spinning at several thousand RPM. As they entered their terminal illness, many drives started to make ominous groaning noises, but it is a long time since I last heard that (touch wood), and I have never heard one self-destruct in the awesome way that some original IBM Deskstar or ‘Deathstar’ models did.
The only crashing sound that I have heard come from a Mac was the smashing of glass (in lieu of the usual startup chime of those days) to indicate a boot problem in some Classic models. And that was a very long time ago.
So the computing sense of crash has no onomatopœa, and as a figurative use it is gross hyperbole compared to almost every other sense of the word. The term is also far too widely encompassing to be meaningful.
These days Macs very seldom (grasping wood even harder) fully crash. The most severe thing that can happen is a kernel panic, which is more the controlled resignation of OS X being unable to cope with what is being thrown at it.
This is a far cry from my early computing days, when omitting a semi-colon in some Algol code returned you a trolley piled with fanfold printout, a core dump, in which the complete contents of core memory were vomited out onto paper. It is as well that old mainframes had less core memory than we now get in Level 1 cache. Now a kernel panic can still generate a core dump, but this is discreetly tucked away in a file instead.
App unexpectedly quits
The great majority of Mac crashes now are less of an explosive fragmentation of glass, and more of a brief and feeble whimper, as an app ‘unexpectedly quits’. For once we do not have to contend with hyperbole, but euphemism. In truth the bloody app vanishes along with all the hard work of the last couple of hours, pausing only to post some irritating alert which tells us nothing we don’t already know and bitterly regret.
Although I am sure that, at the time, we can think of many more appropriate terms, describing this as an application crash is more precise, but far from apposite or cathartic. A better replacement might be an eponym, chosen by popular vote, and for my part I nominate ‘googleearth’.
The other serious misadventure is when a crashing app (or other process) manages to lock your Mac in some way. This might freeze the cursor, so no matter what you try, you cannot regain control unless you force a restart. However it comes in different grades, according to whether you can watch the clock still ticking your precious time away, or everything appears locked. This is also known as ‘hanging’, and is sometimes the fate that you might wish of the developer responsible.
As this is the equivalent of a reduced level of consciousness, we could improve precision by adopting the same approach as in medicine, with an analogue of the Glasgow Coma Scale (GCS). It is quite untrue that this scale was pioneered in Glasgow because of the exceptionally good supply of patients with a wide range of values.
In medicine, GCS is a simple score derived by summing numerical gradings given for the eyes, verbal ability, and motor responses. So we could perhaps factor in pointer control, clock function, and the integrity of display contents instead.
Episodes in which the cursor changes into a spinning beachball are not usually termed crashes, but may as well be. We also need to separate them from incongruous images of well-oiled bronzed bodies on the Californian coast.
I therefore propose that we abandon the term crash in respect of computers. Instead of a fragmented system of ad hoc words, it would probably be simplest to use
- panic for a kernel panic;
- flap when an app unexpectedly quits;
- fright for a freeze, lock, or hang;
- fluster for the spinning beachball.
This leaves alarm and funk reserved for future use, to describe novel crashes that are currently under development.
Does anyone fancy drafting an RFC document?