Yes, I saw that recently.. so it is the butt-dialing case...

I have a few Echo devices in my home... I do have to admit they often mistake unrelated things as wake words... Phrases with "relax" and "ask" in them seem to get confused quite often with "Alexa". We'll be talking at home and periodically Alexa will chime in with some kind of random bit, usually an error, but sometimes she'll start playing some random song.

I do wish Amazon would have some better choices for wake words to avoid false wakeup. I get that the wake has to be implemented in the device, which is hard, but they could also some longer phrase needed after wake to be accepted as a valid command.

Right now we have:
"Amazon" and "Computer"- too likely in normal household conversation for technical people who shop at Amazon. Cute novelty but impractical.
"echo" - too short at 2 syllables, I haven't tested but I assume it is easily confused with "I'll go" and similar
"Alexa" - already having problems with this one.

I could see a simple fix being requiring everything to start with "Alexa please" let the hardware wake on the "alexa" and ignore anything that doesn't immediately follow with please in the processing. You could even make it a customizable add-on word, which would prevent the "TV commercial triggering Alexa" problem as you'd effectively have your own "password" after the wake word.

Part of me wonders what really happened here...

An Alexa device can make phone calls if set up for it, and they apparently had this person's phone number imported into their Alexa contacts, so they clearly had that feature configured.

So was this just a case of Alexa making a phone call, without "Daielle" being aware of it? If the call was to a google voice number or similar service, it would just recorded the unanswered call and emailed them. This case would be just a speech activated equivalent of butt-dialing, paired with a google-voice emailed voicemail twist.

Or was this a case where Amazon generated audio files and then emailed them to some random person out of the contact list?

I kind kind of understand how the first case could happen by accident, but still be disconcerting. However, the second case would be very disconcerting.

Given that Amazon is offering to de-provision the communications feature for them, I'm inclined to think this was a "butt dialing" incident, but I'd love to hear some actual details to confirm one way or the other. Clearly the title of the news article is designed to make you think it was the second case, where Alexa recorded the call, not google voice, but there's a lot of vagueness here that makes it unclear.

Interesting. My cable company puts up the same guide-box on the bottom of the screen every time I change channels, with the same, static header bar. I've had a Panasonic TH-42PX80U Plasma for 4 years (almost 5), with no burn-in issues. I play video games on it too.

I have the pixel-orbiter on, and I do use the screen wiper once a year or so, so that may be some help. Regardless, I find it strange that the box is up long enough to result in image retention. AFAIK, displaying the black box in the same spot frequently shouldn't matter unless it's up there a very large percentage of the time the set is used (i.e.: way more than 10% of the time.).

I know older sets are more vulnerable, but it doesn't sound like yours should be old enough to lack things like a pixel orbiter.

I couldn't agree more.. 65 feet seems really close, probably a bit on the unsafe side, and closer than needed for photography of the spill.

And let's face it, if you can't get a fantastic shot of something large (i.e.: a giant oil spill, not a flea) from 100 feet away, you're either badly equipped or lacking in talent. Getting closer will not help you much.

Of course, we're quick to assume this is impinging on freedom of the press, but the last thing we need is some idiot reporter getting his boat hung up in the boom and damaging it because he was 5' away and a big wave pushed his boat into it.

Agreed, light-pipe type skylights are not new, I have 2 solatubes in my home.

It looks like the big difference with this device is the collection head. Rather than a fixed dome with prism ridges in ti acting as reflector, this new device has an moving sun-tracking parabolic reflector. Looking at it, there's a small photovoltaic cell on the center of the reflector, which presumably powers the motors that make the reflector move to track the sun.

Of course, sun-tracking systems aren't new either, but they're mostly used in power generation systems. I've not seen a sun-tracker applied to a light pipe before.

The parabolic reflector is also not new, but also I've not seen it applied to a light pipe before. It has a much larger solar collection area than your standard dome, so it should bring a lot more light into the tube, at least as long as it stays aimed at the sun. Without a tracking system, this head would produce a lot of light at its peak, and very little light the rest of the day. Domes provide mediocre peak collection, but sustain that mediocre performance across most of the day without any kind of re-aiming, making them effective enough to be useful and cheap.

It would be interesting to see how the extra cost of the sun-tracking head compares to the cost of just installing enough additional dome-head collectors to get the same light output.

True the architect has no rights over the building, but he didn't create that, he created a design and drawings used to create the building from. The question is does he have rights over the drawings? (i.e.: can he use them again, selling to another client because he retained ownership and licensed you a copy, or can he not, because he transferred the ownership of the design to you?). Most don't transfer the copyright of the design to you (ie: you can't make more copies of the drawings and sell them, but he can).

Another common example that exists in the real world is professional photography. You hire a photographer to take pictures at a wedding, paying him for his time to do so. However, you don't own the rights to the photographs he's taken. You can buy copies from him, but he still owns the copyright. As a result, you cannot legally make more copies of those photographs. Photographers generally want additional fees to transfer the copyright of a photo, because it means: 1) you can get copies made elsewhere cutting them out of that profit stream and 2) they can't use it in their portfolio of existing work they show to potential new customers.

Most software consulting is done on a "non-exclusive, unlimited royalty-free license" basis. You have unlimited use of the source code created for you, and the programmer generally retains rights to use that code again in other projects. Most programmers have a lot of "stock code" they've already written, and grab chunks of it to throw into their projects so they're not constantly reinventing the wheel for simple stuff. Some of the new code they write for you is likely to end up as snippets added to that collection. Take a snippet, leave a snippet.

You can set up contracts that transfer 100% ownership, but that's going to increase their billed hours and your cost (because they won't be able to use any of their existing snippets and must create everything from scratch). But if that's what you want, most are willing to set that up. However, most customers want a lower cost so the typical consult is not written up that way.

The moral of the story is, if you want specific terms, make sure they're in the contract you sign.

Well, statisical significance depends a lot on sample size. You can't just say 60% isn't significant relative to the 50% chance of random guessing without knowing how many tests they did.

A 1% deviation from random can be significant if the number of data points measured are absurdly large. ie: if 10 million coin flips come up 51% heads, you have a statistically significant difference, because that 1% still represents a LOT of extra tests that came up heads. However, getting 51% heads in 100 tests isn't significant, because it's only 1 extra head. For that matter, 100% of 2 tests is also insignificant, despite the huge deviation from the norm, the sample size is too small to mean anything.

Looking at the study they do appear to have p values in the statistics in the study.

For "study 2" they quote:
[M = .62, SD = .12; t(23) = 4.91, p.001, r = .72] a

With that p, it seems pretty significant. (unless they fudged the p.. I couldn't find their sample size.).

Do you see anything in those numbers to suggest otherwise?

check your print screen button.

The issue is that the network is oversold to the point of instability, and it seems to be affecting their entire network.

I've had Cingular/AT&T for 5 years now, and I never had any noticeable calling problems up until this year. Over the past 3 months I've had numerous dropped calls and bad reception problems while talking on the phone (handfree, yes this is legal in my state) on the way home from work. This is the same route I've been driving for all 5 years.

I wouldn't be too surprised to have problems in places I've never tried before, but in the same places I've always used my phone in? There's only a few possible causes there.. Overload, hardware failure, and botched reconfiguration are what pop to my mind. Given that we know bandwidth use is rising, and AT&T is cutting back on large equipment purchases, I strongly suspect overload.

Verizon will send you a SIM for your jail-broken iPhone??? Really? Do you know anyone that has done this?

All US versions of the iPhone are GSM / HSDPA based devices, and Verizon's network is CDMA / EVDO. These are fundamentally *VERY* different technologies, and the radio for one can't be re-used on the other. The iPhone can't be used on the Verizon network, regardless of what you do with the SIM, because it has the wrong radio.

There is a CDMA based iPhone out there, but it is designed to work in China and uses differet frequencies than Verizon does (and it doesn't have the right hardware to support the US frequencies, so this can't used either).

Now, t-mobile is another story, their network is GSM based... but Verizon is a no-go without a completely different radio system in the phone. (ie: a new, different model of iPhone is required.)

Well, that's the thing about antitrust. It doesn't really matter what you've told the customer, or what the customer has agreed to. Not telling consumers would be fraud. Antitrust is all about preventing a competitive market.

i.e.: if multiple manufacturers agree to a price-fixing arrangement for some product to raise prices across the market, it doesn't matter what they tell the consumers. They are still illegally colluding to prevent competition in a market.

However, I don't clearly see how this kind of market-impairment argument is likely to succeed in this case. It all boils down to:

Does Apple have the right to modify a product in a way that makes it only operate on one carrier that they have an agreement with, when those modifications serve no other purpose beyond carrier lock-out?

Clearly this does have an impact on competition in the wireless market, but is it a big enough impact to claim that it's restraining trade? Maybe, but it seems a bit of a stretch. These laws are primarily intended to prevent agreements that end up broadly rigging the market, not exclusivity agreements.

Fair enough, I wasn't really discussing the validity of the case, I was merely pointing out you don't need a monopoly to create an anti-trust lawsuit.

However, this does go beyond just an exclusive distribution arrangement. Here you're locking a product, which by design can be used on any GSM cellular network, and limiting it's use by the consumer to a specific carrier.

This would be equivalent to a music CD that has features intentionally added so it can only be played on Toshiba CD players, despite the fact that the disc is otherwise redbook compatible.

Your distribution example limits where a consumer can buy a product, but the reality here is they're also limiting where a consumer can use the product as well.

I still don't think it's likely to prevail, but it is a bit less simple than distribution.

And while this cell-carrier agreements are common, that doesn't mean they're unquestionably legal.