Introducing human.io
joshua.schachter.org
joshua.schachter.org
But I wish there were a specific example of what it does.
Something that starts with, "Let's say you run a site and need [x]. Here's how human.io can help you with that..."
Otherwise it's too much work for me to try and figure out what it does, or how I'd use it. Even though I'm sure it's great - because Josh made it.
(I know that's not what it does. But just illustrating my point. Don't make me, the user, figure it out. Just tell me.)
I have been playing the Photohunt and am currently in the lead - it is very fun! Thanks for building the Human API - I have many great ideas using this and have been taking notes to help improve the process. Contact me Hacker @ myusername ORG
It's still not clear to me whether the Human.io users are tagging photos you send them or taking new photos or what.
Everything is generated by the client. The tasks, the photos, the voting, are all happening in the human.io client. The rest is just driven around by the script. They write to a database, and the web page is just pulling from that database.
This would be easier for you to see if you installed the human.io client.
(FWIW it took a few YEARS before I was able to explain tagging and group behavior on delicious cogently. This process is not surprising.)
So, yes, it's a bit like mechanical turk. Hopefully some of the usage will be driven by participatory urges rather than getting 5c per transaction (though we will probably empower that sort of payment, and more. Offers etc.)
"Human.io lets you to script with humans as easily as you would script with software. It does for the digital economy what the assembly line did for the industrial economy.
The software allows you to dynamically assemble ad hoc armies of virtual workers, who are then assigned to complete tasks in the real world using the sensors and input devices on their phones: GPS, camera, microphone, and so on.
Let's say you want to X. Human.io lets you..."
Still not perfect, but I think it's a little more clear than what you have.
That was our thinking when we first started working on human.io. If GMaps lets you invoke methods on a map, and Twilio on a phone, what would it mean to invoke methods on a person?
I think it's important to have that first, otherwise you are burying the lead. Although I can see why you wouldn't want to prematurely niche yourself, I don't think the pitch about engaging your users is especially compelling.
"what would it mean to invoke methods on a person?"
I mean ultimately you want to be selling the benefits and not the features, so I think being able to communicate the implications of that statement are much more important than being able to come up with a snappy analogy for the statement itself.
If you're really committed to staying neutral on usage though, we already have the term HCI for human-computer interaction. So maybe you could describe it as an HCI programming language or something.
Sample human API tasks: 1. photo scavenger hunt 2. parking garage availability 3. male to female ratio in a club 4. collect blind auction bids
(V)(°,,°)(V)I'm imagining a headline like "Things that used to be really really hard, but are easy with Human.io" and then pictures and captions describing the use cases.
So if I were to give advice, I would say to have coffee with PG and give him your pitch. He's magical. Regardless, grats on the launch!
Having been to more than a few demo days myself, I think this is a much broader thing than YC cos tend to do.
The problem is that what we have built is very, very broad, so narrowing it is difficult.
The platform lets you send people little bundles of UI and have them interact. That's incredibly broad and vague!
So here are some examples: You could get them to take pictures of all the storefronts in a town. You could send them realtime surveys to go along with a television show. You could have them rate sessions at a conference. You could have someone be notified when they walk onto a car dealership and offer them the chance to look up car prices.
Explaining this with one line has been very difficult so far. My favorites so far are: 1) Turn a passive audience into engaged particpants. 2) Turn your audience int your army.
That's marketing speak.
How about: "Let's say you want to build a website that shows the different rates for parking garages around San Francisco, but you don't have the pricing data. Human.io can help you get hoards of strangers to run around the city, snapping pictures of pricing signs at garages, and uploading the pictures to you. Viola, now you have the data."
Is that vaguely what human.io can do?
You could do all sorts of things. How do I wrap it up into one sentence without marketing speak?
Edit to add: "a Mechanical Turk for tasks not easily done at a computer" isn't a bad description either.
...ask users to vote on them using human.io
humans = init({ count:100, gender:male });
task_navigate = task({ type:navigate, uri:{latitude:123, longitude:456} });
task_photograph = task({ type:photograph });
for (human in humans) {
human->task(task_navigate);
human->task(task_photograph);
}
This is a very real problem for the people that have it. Imagine you have a following of hundreds of thousands of people (like joshu actually does). You could do a lot to help yourself and others simply by posting one tweet. But then you have to deal with lots of replies from different people in different formats, keep track of it over time, set up consistent systems for recognition and reward, etc.
human.io turns all of that into an api like process. But you have to explain that idea to a very large number of people, when only a tiny percentage of people even understand what the problem is like. I can understand how this would be difficult to "explain". It seems like the problem isn't so much the message, but the fact that such a specific problem has to be explained to a wide audience.
That's why I keep using the word "audience" - it is the traditional word for this kind of relationship, I think.
Does that make sense?
It seems unrealistic for us to get them all to use the human.io app.
Take his photo scavenger hunt, if that was the only thing this thing was good for it would be a great tool for any major brand.
That's a good idea, thanks. I'll move the section up higher.
We're trying to tell two stories on that page. The developer story: an easy way to write simple apps and target them at specific people via some criteria. The user story: join interesting activities by companies and developers you enjoy or want to help out.
A couple of crazy ideas would be to use this to perhaps improve upon that niche that Flickr and Twitter are sometimes used for, which is crowd sourcing collection of digital media for some event (like a concert, conference or news event). Instead of showing live tweets as a screensaver at a conference, show images collected via human.io.
I also think there is a huge potential for creating interactive narratives and even games.
When I try to explain the idea to people, though, the question is always "what is in it for the worker?". With delicious, the creation of a good folksonomy that added value to the ecosystem was simply a side effect of the selfish benefit of keeping track of your own bookmarks. That's why I think for human.io to go viral to the point where you have a large and engaged workforce, you need to find its "win/win" proposition. It could simply be the game aspect (see games with a purpose) or maybe a give a little/get a little dynamic.
Great stuff...can't wait to see how this one evolves.
Your ideas (conference photos, narratives, etc) are really good ones. Want to prototype some?
I'm not sure that viral is necessary. Delicious was never viral in any way. Just useful.
("turksonomy" - hee)
Some way for workers to see value besides karma or a pay rate of $3/hour is what I haven't figured out yet.
Viral in the classic sense might not have described delicious, but I will say that I had a much richer and more interesting and more professionally useful social network of followers and followed people on delicious than any network since...Twitter is the heir for this kind of network-filtered firehouse of interesting stuff, and Quora almost could get there, but delicious had more impact per byte than either of those. I always considered it somewhat viral in the sense that I had maybe a dozen coworker colleagues I followed plus a smattering of random people who turned out to essentially be high-signal, curating SMEs for various interest areas I needed.
I am working on prototyping ideas as time allows, but still working though the examples so far.
My first thought exactly. This could be epic for so many things that are not practical today. Maybe the incentive could just be: To get the data, you have to participate in creating it.
Would love a "how long are the lines at the nightclubs" task
Edit:
Maybe it's worth looking at how private torrent sites do this, there has to be an incentive to seed or it doesn't work. There's basically a ratio that can't drop too low, or a point system where you start >0. So you have to give X to get Y. Maybe let the users pay for "points".
Do all the end users, the "humans" have to have the human.io mobile app in order to run "apps"? Is there a way to embed the human.io app/task UI somewhere else?
Is the intention to mostly have lots of people use the human.io mobile apps? Or is this going to be more API based where developers can embed human.io apps into other places?
Is the photo scavenger hunt example an app or a task or is it an app with a single task?
For what it's worth, people coming in to do the photohunt app are doing other people's tasks, too.
We want to start with getting people to use the mobile apps. We are looking at other targets in the future (browsers already work. televisions? cars?)
The scavenger hunt is an app with a single task. Apps are really a developer notion - users don't see apps in the UI currently, only tasks.
This may change in the future; inside the one task are several subtasks: suggest a hunt, rate a photo, take a photo. and the "take a photo" part lets you mark a hunt as inappropriate.
In the future these could be four or five simple tasks once you have accepted the larger "task."
I worry that if we do micropayments for microtasks we'll end up in a very specific pigeonhole that isn't insanely interesting.
With this, they could potentially send out notifications to just their audience when participation would be useful. Completing tasks could also be a micropayment allowing the developer to give the user some sort of reward (like unlocking bonus features).
Hit me with some better ideas?
Human.io provides a robust platform for creating mobile activities for people to perform in real life. Quickly create interactive apps that take full advantage of mobile GPS, video, and photo capabilities, promoting rich responses to surveys, games, or contests. Plus, using our integral filtering API, you decide exactly who gets to see and run your app. Push out your app to people based on location, demographics, or your own custom filters.
The UI on the android app could use a little work though. I'm sure you'll get to that soon enough. Nice job!
What's specifically wrong with the android app, so I can make sure Paul & co see them?
we run a small organization that picks fruit growing on trees all over atlanta and donates it to local homeless shelters. we have many fruit picking events that are logistically tough to organize as the trees are all over the city and in all different stages of ripening. if we had a way to break it down like: "you go check this one tree (here's the address), you go check that one, etc, and we'll build the pick around that" it'd be much much easier on us and way more engaging with folks (people love what we're doing, but we have a limited number of events and spaces within those events).
For example: Say I need something from the county records of some place far from me but I need them quickly. I place a human.io call to pay $50 to whomever takes a picture of a series of documents first. That could be huge.
Think of human.io as more of the platform you might build one of these on. We're not reselling a labor pool, we're making it possible to stay engaged with one, especially for folks that have large audiences that aren't particularly engaged (newspapers, etc.)
We're aiming at a different use-case though. Lots of people doing short one-off tasks around opinion and circumstance rather than paid specific tasks with a known, trained workforce.
Imagine if the Chicago Tribune wanted to build data-driven apps from people on the streets downtown. Just like if they wanted to do telephony apps they'd have to spend $300k getting Asterix wired up properly and Twilio makes it work out of a Sinatra app; similarly, my sense is, Human.io is trying to do that for people-driven apps.
(In my parlance, publishers are the people with audience, developers are the people who code, and audience are the people that use the phone. In the wacky future, they may only be indirectly connected.)
Also, I think the emphasis on "humans" in your messaging obscures the point a little. You're not so much programming humans are you are transforming them into sensors.
In that sense, a little surprised the words "eyes and ears" appear nowhere on your front page.
The API sounds good but I may not actually try it until the UI is more compelling.
But in my opinion:
>> It allows you to script with humans as easily as you would script with software.
Is a disgusting phrase.
Hmm. It allows you to get people to do a sequence of actions (provided the right motivation.) What would express it better?
We built it first. Now we're figuring out what it can do, exactly :)