I wish I was surprised that these platforms are moving more and more toward a sharecropper existence for developers. The mobile app ecosystems for iOS and Android are bad enough; you have very little control, but at least you know the terms when you sign up and if people buy/use your software, you get paid something for it (if you choose to put a price on it). It sounds like for this market, Amazon pays you whenever and whatever they feel like, down to and including "never" and "nothing".
Platforms should be utilities. We can't really do without them, and there should not be too many of them for practical reasons. But use them, and we risk giving one or few players all the power. Also platforms give rise to internally regulated markets, and it's one thing if the government regulates a market, but yet another if a company regulates it.
I'm guessing we need special economic rules for platforms, or governments should own them.
One of the worst aspects of this, to me, is when the platform company affords its own apps access to APIs apps from the platform aren't allowed.
As an example consider Apple Pay Cash. It grabs anything you send in the messages app with a dollar sign, and converts it to glittery text that prompts you to pay whoever you're talking to.
Aside from the question of whether that feature is useful to you, it's such a bullshit, cheating way to undercut Square Cash and Venmo.
They're exploiting their position to make an unfair playing ground. Feels so gross to me.
The deck is stacked at every turn, which is dangerous. The sad thing is that I've always had sort of a belief that "more open wins in the end", but that hasn't really been borne out in the mobile space. Apple and Google have built walled gardens where they always have the advantage in any space they want to operate in (even moreso than Microsoft during its rise to dominance, though they're slightly more subtle about how they wield that advantage to kill competitors). It's pretty much a battle of good vs evil all over again, and evil is winning. It's alarming that new markets are even worse. I haven't looked into the voice apps market for alternatives to Alexa, but I assume if Amazon thinks they can get away with acting this way, the rest of them are the same or worse.
I wonder if government ownership solves the problem, and whether it might create new ones (lack of innovation being a big one). But, there is definitely a major problem, and I'm hopeful that an open internet is able to solve it.
There's the problem of API access; you still can't make a mobile web app as good as a native mobile app and it may always be playing catch-up (both because of policies at Apple and Google and because of the nature of chasing new features from outside). But, at least it really is "our" platform (if we can keep it). With WebAssembly enabling development in any language (eventually) and easing porting from native to web, things are looking up, but it's still not an even playing field.
So...on optimistic days, I think the web will win (again), and will prevent the mobile OS walled gardens from owning our future. On pessimistic days, I read an article about a new platform for apps that is even worse for developers than the App Store or Play Store.
I guess my action plan is: Help the web win and hope the app stores die.
I agree mostly. Lack of innovation could be solved by letting the government own and define the platform, not build it; instead they can use contractors, which could be large tech companies. Also, universities can help design the requirements.
Note that the definition of platform is broader than OSes. For example, Uber and Booking.com are also platforms (and they have similar problems as OS platforms).
I think developers and really anyone dealing with Amazon needs to understand that there are no 'win-win' business relationships with Amazon. If you win, Amazon will try to take it away from you eventually.
I started to build a thing (maybe an OpenAlexa) for this reason (glider5.com) - the goal was to eventually hook it up to STT and TTS and have it answer queries, with all the code being open. Right now it kind of looks like a dysfunctional wolframalpha - but where anyone can write the code to answer queries.
Not sure whether to just kill it or do something else with it, but it was a fun idea.
I've been playing with Mycroft and Jasper before for the purpose of having an assistant running on a PI but I stopped for different reasons. There is a need for a OpenAlexa assistant out there for privacy aware customers. My gf is pretty excited with the stories emerging around her friends who got Alexa. I'm pretty sure this will expand as an assistant becomes normal. But I don't want to have a Amazon employee at home.
Alexa skills are weird in that Amazon is supposedly selling lots of Echos but our skill has something like 500 active users when our mobile apps are in the 1M+ active range.
We have no idea why skill usage is low. It’s a really good skill. The major complaint seems to be having to use our app name to activate it but we have no control over that.
I use my Alexa's (3 of them around the house so far) many times a day, but the thing is, there are only so many things I want to speak out loud. It's exacerbated by having to go in "blind" - I'm not going to remember the incantations to get a response out of dozens of skills, so they - including the name - need to fall very close to something I'd naturally say.
Overall, I mostly rely on the smart home stuff, basic searches, music, and one extra skill for local travel.
I can't see myself opting to use skills over my phone, tablet or laptop for all that many other things. The above is enough to make me enjoy having them accessible everywhere in the house, so I'll probably augment my existing ones with a few Dot's.
It also doesn't help that there really is a lot of junk skills around. E.g. finding one that reliably provide bus times near me and that has a sufficiently good interface was a pain. There might very well be other skills out there that do a better job, but I simply don't have time to spend going through it. Good curated collections would help a lot.
I think Amazon (or Google) needs to solve the problem of skill discovery. I would love to be able to ask Alexa, "What do you know about X" and it could respond with a list of skills or categories of skills so I could at least know where to look.
Even better if I could connect Alexa to a Fire Stick and have it visually display results of a skill search to help me navigate and drill down.
I think there are a lot of factors at play. I have had an Amazon Tap for about a year, and 99% of the usage is just to play music. And I work in technology. I have used the Alexa app on my phone probably 3-5 times since first setting it up. At that time, I installed a couple of skills:
Pandora
Jeopardy (which I played often for a while, but then stopped)
Some Bed Time story skill, which is awesome and hilarious and I should know what it is so I can recommend it.
Nest, which never worked because I guess you have to use Heat/Cool mode on your thermostat which is dumb.
I think probably:
1) I installed more skills than the average user on initial setup (expected median: 0)
2) Most users open the app at a similar frequency to me.
3) The primary use of the device for most users is music.
4) Some people don't use them at all after the first week.
As far as development of skills, I know at some point they were offering "Amazon Swag (apparel)" in exchange for uploading your first Skill. Come on people, I've got mouths to feed here!
>> Nest, which never worked because I guess you have to use Heat/Cool mode on your thermostat which is dumb.
I have mine on "Heat" mode right now and it works fine. I think the problem is the name of your Nest. Mine was originally "Dining Room" (I only have one thermostat, so that was a dumb name, but recommended by Nest). I changed it to "Home" since I got the Echo, and it works fine.
Yep, my usage is about the same. I would say 99% of the time I am just using it to play music and then the other 1% is just asking it simple things like the weather or what time it is.
I installed a couple goofy skills but rarely use them. I did set it up with a custom skill to control my Roku by voice, but I don't think I have used it since I got it working.
I found it to be a game changer with respect to smart home.
Most of my stuff was underutilized, i.e., Nest, Wemo, Harmony Hub. Since I got my Echos, I've accelerated my timeline to put more smart switches throughout the house.
My favorite routine with the dark winter mornings is to use a single voice command to turn on my main floor lights and tell me the temperature outside as I groggily take my dogs downstairs to go outside. The temperature announcement lets me know whether my dogs need their coats on. It was especially useful today, as it was -23C this morning.
Hey, can I ask which smart switches are you using? I have been searching for a smart switch that:
1) Can have up to 3 individual switches in a single outlet.
2) Can work with just a single hot wire (no neutral).
3) Can be queried about it's status (on/off) back to the gateway.
There are a bunch of Chinese ones that meet 1 and 2. But I haven't found 3 so far.
I currently have 3 WeMo on/off light switches and a couple of plug in light switches.
I am planning to get some Z-wave light switches for my 3-ways (Zooz seems to make the switches that work the way I want).
1) Can you clarify this question - I'm not sure I understand what you're talking about. By outlet are you referring to a junction box or an electrical outlet (plug)?
2) Wemos require hot, neutral and ground. AFAIK, most smart switches at the very least want a neutral wire.
3) Wemos don't need a gateway (wifi), but they know their status.
If I could do it all over again, I probably would have skipped Wemo and went with Zwave instead.
The neutral wire requirement is really a killer for me, as I'm not going to rewire me entire house. There are some switches that only require the hot wire (they pass a very small load to the light, but not enough to lit up even very efficient leds).
As of the gateway I don't really care the interface. I'm prepared to write code myself or whatever is needed.
Ok, now that I see a photo. I don't know if there are any vendors that have single junction box switches that control three lights. In general though, the smart switches tend to be deep and don't leave a lot of room in the junction box.
Are you sure you don't have a neutral wire in the junction box? In my junction boxes, the neutral wires weren't used for the old school switches but are capped in the back of the junction box.
Regarding the interface, the Wemos are not very reliable because they are wifi. I used to use X10 and they were way more robust than the Wemos. You might want to check X10 if you're going to code yourself. The hardware tends to be a lot cheaper than the newer switches.
Yeah, I would be all over that if my home was newer and had wiring to support it. Right now I don't think there are many options with the way older homes were wired.
The installation of skills doesn't seems as intuitive as installing an 'App' on your mobile phone, therefore there seems to be a technological barrier.
Not sure if this is new but I recently installed a skill via voice on Christmas. I'm not usually one to "test" Alexa but I said "Alexa, when is the next Bainbridge Island ferry?" just to see if she could query it.
She said, basically, "The Washington Ferry skill can probably help you answer that. Should I install it?" I replied "Yes" and she took a moment to load the skill and then also automatically answer my original question.
hbosch seems to describe the ability to install skills by voice, if that's not what you had in mind is it possible to install a skill that makes installing skills easier?
> The major complaint seems to be having to use our app name
A minor sidenote: Picking a good appname is vastly more important for Alexa than it is for typical mobile apps. It needs to be something people are not ashamed to say out loud.
Overall it seems that time spent using skills/apps on Echo devices is a relatively low proportion of the total time spent using the device (relative to the proportion of time people are using apps on their mobile devices, that is). Time could
certainly change this, however
Any written tutorials? I've never understand the preference for video for coding tutorials. (It's actually the speaking more than the video I dislike. Or rather it's the enforced pacing that time-based media impose on you.)
Some people prefer classrooms rather then books. Videos work like that, except you can speed up or pause the video. You can also easily skip around to the topic that interests you, and you'll easily see how long it takes to learn.
Videos also show you how it works and how it's done. Books cannot do that by mere fact that they are static pages. They can describe and have screenshots.
Hopefully this gives you a better understanding why some people prefer video over text for everything.
To be honest skills have no easy monetization path if at all.
I’m not even sure skills really need to exist for anything that only uses the Alexa UI. For IOT devices it makes sense . The big issue is that you cant purchase skills and to get to using a skill is time consuming.
I made a skill called history facts and I don’t even use it. I never use skills at all.
Most of our use of Alexa revolves around skills. But they are all around smart home products we have like Wemo, Harmony Hub, etc.
The bugbear I have (being a Canadian user) is the region lockouts. A lot of existing skills aren't ticked for English (Canada) support, so we're stuck waiting for skills to show up.
I contacted some product companies for some skills I was waiting for, and it didn't even seem like companies were notified of the region setting or the recent Canadian release.
On a side note - navigating and consuming content on cnet's site has to be one of the most painful experiences on the web. Slow and ugly only begin to describe it.
The article mentions my favorite Alexa skill: Jeopardy. It's a testament to how well the hardware + voice recognition works that you can reasonably play through 12 trivia questions daily with only minor hiccups.
It's not multiple choice, you just answer with whatever phrase makes sense and it mostly just _works_ (which still strikes me as fairly amazing).
That sounds much better than the other trivia games I have tried, which have been pretty awful, and reflect poorly on Alexa. They seem to fit a 5 question, multiple choice template.
Wolfram Alpha debuted in 2009. Just imagine what they could have been if they'd opened up "skill" creation to the public.
Given that devices equivalent to Alexa or Google Home could pretty much be launched on Kickstarter, is the ecosystem large enough to allow an open-source or more public version of the AI assistant? Is it needed?
Wolfram Alpha actually has a pretty significant collection of “widgets” (http://www.wolframalpha.com/widgets/). These are commonly embedded on math and science related pages. The more clever widgets take full advantage of Wolfram Alpha’s NLP abilities.
As to why they didn’t have Alexa-style voice skills, I’m not sure speech recognition is a core competency of the company.
It was a Dev from Ibm who used to develop for Alexa, but with its' unpredictable payouts , he has spent more time on the Google Home platform which allowed him to advertise his books through Google AdSense.
When I created my skill I created it for myself. The free t-shirt for being one of the first 1000 skills was a bonus.
Even though it has a glaring bug, it still gets more traffic than any of my other personal projects. And Amazon (usually once a year around December) offers $100 credit which I can apply to my AWS account for the traffic it incurs (which still costs me nothing).
The article says that the Alexa economy is increasingly important to Amazon, yet if you go on the developer boards, most of the complaints from last year are still there. The flaws in the system keep serious developers away. Those on Hacker News might remember this conversation:
On this last point, it is frustrating that we have not seen more competition. Most startups lack the resources to build their own voice system, so they want to build on the back of service being offered by Amazon or Apple or ...
Apple promised to open Siri, but so far you can't build Enterprise apps with Siri.
There is interest among managers. Among the ideas that business leaders have asked me about:
1. An Amazon Alexa in every hotel room, ready to answer guests questions
2. A salesperson leaves a meeting where they made a sale. They get in their car and drive home. They talk to their ______ device and all of the data is sent to Salesforce, the app creating a new Opportunity or new Tasks.
3. An Amazon Alexa skill that can reach into the companies Salesforce account and generate a report for high level executives
Of this last one, we had a working demo in early 2016, but we need some extra tools for specifying the phonetic sound of a company name, as Alexa mangles the names of companies such as IBM and CNET.
As to #2, we worked on a text interface for Salesforce in 2015. I wrote about that extensively here:
We then investigated how to do this with voice. At the moment, it seems the best option is to use voice-to-text on an iPhone or Android. It's an open question whether it is best to use IBM Watson's NLP tools or build a specialized tool just for dealing with Salesforce.
For your Salesforce bridge, have you checked out something like Twilio Understand? It helps you build a back-and-forth conversation and makes it easier to capture user responses as structured/templatised data. https://www.twilio.com/understand
Honest question - why would you do this? Without any possibility of direct revenue for any skills you build, what’s the motivation? It seems that all you’d be doing is enriching the Alexa ecosystem out of your own pocket.
