Amazon Takes Pikachu To The Patent Office

theodp writes "On Tuesday, Amazon was awarded a patent for Search Query Autocompletion. From the Summary of the Invention--'For example, if Pokemon toys are currently the best selling or most-frequently-searched-for items within the database, the term POKEMON may be suggested whenever a user enters the letters "PO," even though many hundreds of other items in the database may start with "PO.'" See, Amazon practices the mantra "Gotta catch 'em all" with patents.

Comments

Fine for some things...

[Elvisisdead]

...but not for others. Great for entering URLs you've visited before or text messaging, but suh-ucks in word processing. Thanks, I can write a sentence (or in this case, 1 word) for myself.

Re:Fine for some things...

[pmz]

Great for entering URLs you've visited before or text messaging, but suh-ucks in word processing.

The reason autocomplete sucks for word processing is the constant interruptions in the natural flow of typing. Then, once a person is used to autocomplete, the habits formed totally trash productivity in non-autocomplete environments.

I think the best compromise is the tab-to-complete feature in bash and emacs, for example. It doesn't do anything until the user presses the tab key, and, then, it is pretty natural to begin a new word after a tab.

The Amazon patent, however, is not autocompletion, but smart marketing. By flashing the most popular product name with each character typed, they gain instant attention and better chances at impulse purchases. It's sort of like an electronic version of check-out aisles with all the candy bars and trash magazines leading to the register. ...I think I finally understand, now, why grocery stores don't use the more efficient single-queue/multiple-registers model for check-out. Forcing customers into the horrendously ineffecient mode of standing in multiple lines increases customer exposure to all the crap they put in the "impulse zone." Damn, marketing people are evil.

Innovotive.

[GothChip]

Unlike the other patents this does actually look like an original idea.

Re:Innovotive.

[bob_jenkins]

1. On-the-fly completion was really neat the first time I saw that. Ross Comer implemented it in junior high for has ABSTAR program (absent/tardy) in 1982 (1981?) that he sold to all the Ohio schools. It would on-the-fly guess a student's name (first match in alphabetical order) as you typed it in. Written in QBASIC, I think. He went on to work for Microsoft, on Excel.

2. Autocompletion in most popular order, rather than alphabetical order. Looks new and useful to me. That approach will autocomplete sooner. You could sort your whole index that way. Changes in ranking would reorder high level branches of the index, which is kind of weird, but I think it would still allow updates with good efficiency and concurrency. There's the issue of whether you want the most likely next letter, or the most likely entire completion. I'd have to test both methods to be sure, but my guess is the most likely entire completion is more useable, which is what Amazon patented. Autocompletions that partially but don't entirely match what I want to type sometimes throw me off.

Unless there's prior art on #2, it looks like a valid patent to me.

(I agree that the world would be better off if this, and every other software innovation, wasn't patentable. Patents just hold back progress.)

What a waste of bandwidth

[cryptochrome]

Listen Amazon, your website is slow enough - no need to slow it further by constantly pumping partial queries and results over the net.

Assuming you can get a patent on something as obvious as autocompletion. Whatever happened to not granting patents to the trivial, the almost-identical, and the prior-arted?

Patent is strategic

[binaryDigit]

Amazon doesn't care if the patent can be canceled due to prior art. They'll strong arm other companies and many are bound to not put up a fight. If someone does, and the patent is later invalidated, then the max they'd lose would be to have to pay the original licensee back, I don't believe they'd have to pay any type of penalty on any fees collected. So they basically end up with a interest free loan, IF the thing gets invalidated. Not a bad downside. The way that the current patent system is setup, your much better off trying to patent everything, as even if a large number get punted, you'll probably make good money off the ones that don't (kinda like VC in the boom).

Re:erm.. is this patent G rated?

[geschild]

Good point. So a filter it is. Which makes this functionality next to useless because people will be 'Pissed Off' (pardon the pun) by systems that get their intent wrong most of the time. (Or if it takes typing in an almost complete word before it hits the right one).

One of the reasons people despise clippy is because it is constantly guestimating. badly...

Smells like Marketing

[mobileskimo]

Its not
"Where do you want to go today?"

It's
"Where do we want you to go today?"

Avalanche

[blunte]

If we thought we were seeing too many software/business patents, we're about to really be amazed.

Now we're patenting "features" of software--behaviors even. How about Undo? Oooh, that's worthy of a patent. Or double-click to select a word, triple-click to select a sentence?

Pick any feature of any software system, and it's now fair game for patent. This means of course, in the future you'll have to get a licensing agreement from FubarU.com, the patent holder of the "Undo" feature.

What I wonder though, is it just pure malice that drives these humans to patent things like this? It certainly can't be business sense, since Amazon can't conceivably get any more online retail business by others not being able to use this feature on their retail sites. And it can't just be for license fees, since those may or may not ever come to fruition.

What ever happened to the good old days of insurance fraud, embezzlement, and plain old theft? At least those perpetrators had balls.

A few more moderate points

[dpille]

1) This isn't such a "IE did this first" issue. If you read the claims, the process calls for some more detailed analysis on the suggestion end- for example, culling out null results. It'd be the equivalent of IE not autocompleting to 404's, which we all know it still doesn't do.

2) Prior art from any time after their filing date in 2000 won't matter, so don't worry about what was going on "last year."

3) The examiner clearly considered mere autocompleting- look at the references cited during prosecution. PDA operating instructions are among them, which I imagine contained lots of "this device will complete your word for you."

4) Prior posters seem to be confusing "novelty" with "non-obviousness." I think it's pretty likely Amazon was among the first to use this invention as disclosed, but I'm willing to grant that any reasonable programmer turning his or her mind to this problem would have created a similar solution. But that doesn't mean it really has been done before.

Annoying, But

[praxis]

For the record, this kind of behavior bothers me. But, they really are within the confines of the patent system.

All the prior art examples I've seen posted have been about autocompletion or searching a users previously entered text. They are taking this and expanding it to search the entered text of a group of users, giving the benefit of possible autocompetion of text you may have never typed.

Patents are supposed to do this. They exist so that someone can take someone's idea and exand on it. That's what they are doing. There very well me prior art on *their* idea, but so far all prior art has been on standalone autocomplete.

And now...I should say that this is just plain stupid. I never thought something like this should be patentable, but it is. It's the system's fault, and it needs to be fixed. And although they are within the confines of the system, they are just contributing to it's failings. Of course, that could have the effect of more evidence to its demise and rethinking by providing even more examples of misuse of the system.