Wednesday, February 14, 2018

Good vs. Better at Bad

Joe Cieplinski:

Without knowing where “good” is, anyone can wave either one of these comparisons away and chalk it up to priorities. Some people care more about the sound quality. Some people care more about the smart stuff. Sounds like a toss-up, right?

But there’s a threshold of quality where people consider something “good.” Where the general public—not just a niche of enthusiasts—agrees that a technology has gotten significantly good enough to make it ready for prime time.

We reached the “good” threshold for speakers decades ago. The subcategory of affordable bookshelf speakers got there sometime in the past few years.

But we’re nowhere near “good” yet when it comes to digital assistants.

I like the way his final chart visualizes this.

So yes, other platforms may currently be “better” than Siri. But when none of the platforms is good, what difference does that make, except to a small niche of enthusiasts? By all means, enjoy the Echo if you want to live on the bleeding edge of voice assistants. But don’t try to convince me Apple is doomed in this space[…]

The “doomed” narrative has taken hold because it seemed like Siri started out ahead but has fallen behind and is now facing a network effect. What reason do we have to think that it can or will catch up? Still, it’s far too early to know who will win the race, or whether that’s even the right question to ask. Perhaps at some point they’ll all be good enough that people choose based on other factors. Some would argue that’s what happened with maps, where Google remains ahead, and may even be increasing its lead, but yet Apple Maps is improving in an absolute sense and many people use it successfully.

Update (2018-02-15): Nick Heer:

But I maintain that, even if Amazon and Google aren’t that much closer to a fully assistive software or hardware product, the ways in which Siri frequently fails are unacceptable.

Update (2018-02-23): See also: Matt Birchler.

7 Comments RSS · Twitter

I think the most astute point from Joe's post has to do with how we get from where voice assistants are today to where they need to be in order to be truly revolutionary:

>Because here’s the thing: that level of complexity is not just a matter of gathering more data and training our AI models a little longer. It’s not a matter of third party apps. It’s not a matter of open vs. closed. It’s not a linear progression from where we are today to *that*.

Benedict Evans is another smart tech analyst who has made the same point. Not only is the required level of AI to realize this vision probably many decades away (if it's possible at all), but something like Alexa is not really even a step in that direction. Alexa is a basically voice command line. That's useful for certain things, but it's not step one on the way to some kind of true general AI assistant.

If people could just understand Apple keeps putting out "me too" products for the last several years. Me too products that are rarely better than their competition (some are barely competitive period)....then we wouldn't have to have hand wringing articles like these every time the new Apple device probably isn't that great given it's price and brand prestige.

Seems like there was a time that Apple was rarely first, but was often better or differentiated enough to show strong merit. Now it seems like Apple is never first and rarely better. Who cares really? If you see a device that meets your needs and price point, buy it. If not, don't buy it.

Here's the thing, I do agree that we don't really know what the future might bring. I remember when Roku went into a partnership with Netflix and released the Roku DVP....which had the sole function of playing Netflix. You couldn't even browse the catalog on device, you had to save videos via the Watchlist on a device with a web browser. Only then could you watch any content.

Now Roku is a whole platform that is doing really cool things. Probably the most content agnostic platform given comparisons to Apple TV, Fire, and possibly even Android/Google TV stuff (in order of least to most open....probably). So Apple could grow into something competitive, but I just don't know it it's in the DNA anymore. Big companies have such struggles; yes, even when succeeding. That's how big companies fail....if they aren't careful. Look at Nokia for a very recent example.

I meant Apple could grow the HomePod into something competitive. Look, at worst, seems like the HomePod has competitive sound (better than some, worse than some), but is weak given the non Apple ecosystem support and Siri in particular having some failings. Does that seem fair?

Assuming that's a fair enough comparison, that's a solid building block. The question is "Will Apple put enough resources into sanding out the rough edges or will it basically wilt under neglect like the Apple TV?" Yes, I know the Apple TV is much better now than it was in the first, second, and third gen days....I just meant it seems like Apple could be killing the market with a really cool little TV unit and it just never really went anywhere compared to the other platforms. Shoot, there's many people arguing that Siri just doesn't seem to be getting any better and it's a key integrated selling point of nearly everything Apple is doing! What chance do these adjunct devices have, Apple TV, HomePod, etc. if Siri development is lagging behind its rivals?

Could be wrong. Just a former Apple community inside peering in from the outside.

I find this argument confusing. If sound quality is good enough on all devices, but the quality of the digital assistant is not, and differences there are more obvious, doesn't that make Apple's attempt at competing on sound quality even more misguided?

Lukas, let me put it this way: if you are playing a 2 person game where the first person to get 100 points wins, a score of 80-40 is pretty different from a score of 2-1.

In other words, if Amazon is 80% of the way to creating the ideal virtual assistant and Apple is 40% away, that would be a much bigger problem for Apple than if Amazon is 2% of the way to creating the ideal virtual assistant and Apple is 1% away.

> In other words, if Amazon is 80% of the way to creating the ideal virtual assistant and Apple is 40%
> away, that would be a much bigger problem for Apple than if Amazon is 2% of the way to creating
> the ideal virtual assistant and Apple is 1% away.

As somebody interested in buying a device, though, the opposite is true. Once things are good enough, people stop caring about differences. It's when things are still developing and new that differences are obvious.

In 1986, the difference between a Mac (which was maybe 3% of what a PC should be) and a Windows 1.0 (which was maybe 1% of what a PC should be) was much more relevant than today, where both are kinda good enough, to the point where any differences no longer matter to 99% of people.

If I'm buying a digital assistant now, I don't really care about the sound quality. They all sound okay. I do, however, care very much about how well the actual digital assistant part works, because there, the differences are real and noticeable.

If you think purely about how far Apple has to go to catch up, then yes, making up 1% might be easier than making up 40% (although I think even that is disputable). But I don't really care about that. That's Tim Cook's problem. And it's a pretty big problem, because progress with these digital assistants relies on data, and Apple probably has much less of that than Amazon.

Joe's view seems to be that the Echo's sound quality is not "good enough", though he's a musician so I don't think he's representative of the broader public. I'm more of your view about how most people think of "good enough" sound quality.

But for the long term question of how voice assistants play out, I think Joe's point is very well made. Apple is *not* very far behind in any kind of absolute sense. Not only are they not behind in data as you suggest (they get two billion Siri requests a week), but it's not even clear that the current technological development of voice assistants is the right path to the ideal future AI assistant. It might be a completely orthoganal technology path.

Google, Amazon, and Apple all seem to have the natural language processing problem pretty well supported i.e. understanding spoken words. The problem of figuring out how to handle queries is still very much up in the air. Alexa is trying the "giant if-statement" approach which works really well for certain things and provides a simple way to plug in 3rd party integrations. But there are questions as to just how far it can scale; it's hard to see how they scale that approach from where it is now to Joe's imagined "book a flight" use case.

Apple is trying to be a bit more sophisticated by using a less structured approach, but because they can't make Siri anywhere near as smart as she needs to be to handle general queries, the result is inconsistency, frustration, and a system in which it's much more difficult to plug in 3rd party services.

I think we're in agreement that that Amazon's approach makes for a better product in the short term. But the long term future of voice assistants is very much in the air, and if I had to bet on whether Amazon or Apple would have the superior voice assistant product in 10 years then Amazon's current lead, such that it is, wouldn't make them the obvious favorite.

Leave a Comment