What ideas and projects will take over communities and in turn the world in next 3 years?
I personally think- adtech is very saturated. DSP have become commodities with each company coming up of their own. There is a good chance for innovation in this space.
Potential risk: choosing bad content could cause more conflicts than new connections.
Why not done: just came up with it, also unlikely to be profitable unless you get a research grant for it.
Trigger: this idea is similar, right? https://twitter.com/Aelkus/status/796845589254275072
> but not like that
Customers simply type what they want. GPS locates the customer. Companies submit reverse bids both manually and algorithmically. Customer accepts what suits them best. Yay, money saved, time saved, customer happy.
Specifically, I don't have enough experience building android apps to make it as slick as I would like. Happy to talk with anyone about it. Can build back end, will travel, etc :).
I would love to be able to write somewhere "my backyard aluminium door has problem closing/is stuck" and get quotes from whoever is the indicated person/profession to fix that.
I started learning React Native to do that, but darn I just don't have enough time :(
I would like to see a larger effort for the SV community to expand into some of these communities and help give some of these communities a much-needed economic jolt.
It's time to go to the next level, with universal income and less (much less) working hours, free education and so forth.
The alternative is dark ages + war.
1) Some kind of crawler that goes through YouTube videos indexing them and its subtitles and then when a user searches a word or idiom the web app would show some part of a video that speaks exactly that search term. I know there are other services that does that, but I want something more natural since these services use some kind of text to speech with a robotic voice and not a real life situation.
2) An old school open world MMORPG with modern graphics and a mix of Ultima Online (PVP) and Diablo III (Dungeons) gameplay. Nowadays every mmorpg is a copy of World of Warcraft with a different story. There is no innovation...
EDIT: add another idea :)
Bonus if you can finagle all of this with free AWS credits from their startup incubator offers.
For people like myself who like great art but have very little knowledge about the relationships or even chronology of the various movements, it would be great to have a broad range of artworks representative of the various movements placed on a map. Together with a slider type interface to choose time periods I think you could get a really good idea of how and when the different movements evolved and how they relate to each other. You could easily see for instance that there wasn't much happening in city x in a particular decade but a lot going on in city y. Move forward a decade and while y might be in decline, you can easily see that city z is becoming active - and that the newer works have clearly been influenced by what was happening in city y a decade earlier.
I have a fairly clear idea of how something like this could be monetised, together with how to get the necessary content. Haven't given much thought to how it might be marketed though. You could call it 'Art in Space and Time' (artisat).
I like that it presents a great challenge from a UI perspective - it could be done really well or really badly.
Anyhow, work keeps getting in the way...
If anyone likes the idea, go for it. Get in touch if you'd like to hear a little more - I might even be willing to get involved.
[1] https://minireference.com/static/tutorials/conceptmap.pdf
We've seen it in Pluralsight or FreecodeCamp where they made the paths for their users. Now, if we can turn it around by making it crowd sourced and to a wider skills, that'd be awesome.
But I mostly abandoned it for now, as expii.com seems to be pretty close to what I wanted to have. Also since I finished my studies, my learning changed to be more explorative and need-driven, which didn't fit well into the thinking behind Owleo.
Step 1: My wife and I both download an app.
Step 2: We link our accounts.
Step 3: We each find and read articles in the app, Pocket-style.
When done with an article, swipe right if I think my wife should read it. Swipe left if I don't want to pass it on.
Step 4: Open the queue of articles that my wife thinks I should read. Read them. Swipe right if I want to chat about it with her. Swipe left if I don't.
Step 5: On the weekends, as we're going for our walk, talk about the articles we both found intellectually stimulating.
If anyone starts working on this, email me.
Here is an outline of the technical requirements and how they could be achieved.
https://github.com/andrewt3000/DL4NLP/blob/master/carl.md#co...
I couldn't build it because I a) lack hardware expertise to figure out tricks to navigate without GPS or/and map buildings and b) never figured out a realistic GTM for this.
But what about proof assistants? It struck me that using a proof assistant to guide the theorem prover inside a partial evaluator offers an opportunity to repartition the coding task. The idea of "end user programming" always fails because we cannot sufficiently separate the idea of a program that serves as specification (here is what we want done, its a shame that this specification isn't really runable because the implied algorithm is too inefficient (think about PROLOG, where a naive program may well be logically correct, but backtracks itself an exponential runtime)) and a program that describes a tolerably efficient algorithm. The poor end user has enough trouble with writing code that specifies the computation correctly without worrying about computer-sciencey stuff about how the computer should actually go about computing it.
If only there was some way the person rewriting the code to make it go fast could do so at one remove. He works with the end user programmer first time through, but subsequent revisions by the end user get rewritten automatically...
This was the vision I tried to articulate in https://groups.google.com/d/msg/comp.lang.lisp/5p8oVP5dGbU/V... but my health has failed and I've not got any further forward :-(
A polished, credible implementation of that would mark the beginning of a new marketplace centered on agent services that prey on existing Web companies and tear open their data silos and adtech, one user at a time.
But what ultimately stops me from further consideration is that the level of AI know-how needed to do this is not a commodity at this time - you'd need a full team of top talent, and it probably wouldn't be cheap/fast/customizable/secure enough for users if done now. I think it will happen though, the signs of it happening are gradually emerging, and when it does, watch out!
On the service-end, it would have basically been an Apache Spark instance doing collaborative filtering as data comes in. It seemed like a pretty sweet idea, but I couldn't find the time to work on it.
If this seems like a cool idea, you should definitely steal it and start working on it.
Related to this -- I wrote a blog post about a lot of my random ideas I haven't finished and why you should steal them. Complete list and blog post here: https://kristianfreeman.com/post/steal/
In more detail, an innocuous autonomous droid that reminds me of meetings and takes me to the right location. Intelligent enough to correct for diversions to the wc or coffee machine, and stay out from under foot.
Note that this is partly independent from speech recognition. Perhaps speech could be translated to a string of word alternatives with probabilities, and language understanding gives a hint which alternatives make more sense.
In other words, language understanding could help speech recognition.
This is already done, but probabilities are used to determine the best alternative.
So basically we ask
if p("which" | "Ding dong", "is dead") < p("witch" | "Ding dong", "is dead")
then print("witch")
This generally doesn't give us an understanding of the text.
The main focus of my proposal is something different: a more intuitive «understanding» using neural networks and lexicons. This is somewhat orthogonal to speech recognition.
Then I suggested that this «understanding» could help make speech recognition more reliable by adjusting of some probabilities.
You see, I have an important insight as a Deaf person. Lipreading only offers a small part of what a person could understand by hearing. But I am often still able to understand what a person says by guessing and using the discourse context to fill in the blanks. I am sure that we all do that to some extent, or we wouldn't have homonyms in the language.
This insight tells me that perhaps a disrupting innovation would be a more natural «understanding» of text (in written form) with the help of neural networks and a lexicon.
For me an important use would be a tool to make speech recognition more reliable, for example to automatically subtitle Youtube videos with better accuracy.
Do you mean capturing the thought train on devices and with adaptive feedback mechanism to generate new ideas?
the secondary market here is you search by tree and find fruit/seeds/nuts. A more generic, iPlant would also show vegetables/crops and hook this up to seasonal sales.
I've been noticing the market for local fresh food is pretty big. If you have a system that decentralises the growing it could used as a peer to peer sales or arbitrage where people buy others and re-sell.
As for the iPlantTrees, yes... This is what has been done in my hometown, Melbourne ~ http://melbourneurbanforestvisual.com.au
My single friend tells me of dissapointing dates where the profile photos are years old. She especially dislikes guys who lie about their age to date women in their 20's.
Certain I'd like to talk with you about your strategy and see if we can get any closer than you did the past three times.
I was trying to build a clean static system I could use as a bootstrap to build clean packages from. Ran into trouble with things like coreutils, util-linux, and fakeroot. Making things harder is that I'm building everything against musl libc because glibc doesn't handle static linking well (and also because I don't like glibc).
I'm thinking a better strategy might be to use some other host system with pacman to build a base system of clean static packages, then use pacstrap to get this system up and running.
The 5-star McDonalds may be a really good McDonalds that's very clean and consistent, but the $50/meal ocean-side bar only has 3-stars because the steak is sometimes rare instead of medium-rare. People are rating relative to their expectations of the restaurant, so someone who doesn't share those same expectations will be confused that a dollar cheese burger is 5 stars but a prime rib steak is 3 stars.
Examples:
- if a person only posts 5-star or 1-star reviews, they are polarised and/or punitive. Any and all reviews from such users should be weighted accordingly - 5 stars clearly means "thumbs up, for whatever reason" and accordingly 1 star means "thumbs down, for whatever reason".
- if a person posts a fair selection of review scores, they are likely more thoughtful. Individual review scores from these users are more accurate.
- if a person posts mostly 2/3/4-star reviews, then any 1/5 star review is likely to carry a lot more weight.
To make score manipulation slightly harder, some kind of meta moderation would need to be in place too. Preferably on a per-category basis, so your reviews might have "good karma" on some topics and "bad karma" on others.
1/ So the idea I haven't done is build a hacker search engine. I'm not sure what this would look like, but I'd bet being able to search through source code at high/low levels might be interesting.
2/ Another idea is I noted on the the Amazon/NES classic thread, arbitrage by Amazon/users resulting in instant sellout. Could this idea be used somehow?
dunno, haven't looked at this, will do. Having access to any public source code would be interesting to find ^canonical^ examples. I know where this does't happen, cough stack overflow cough
To see how we built it and future plans, check out https://sourcegraph.com/plan.
Great, will do.
It was called "Google Code Search". 2005-2013. [1]
Actually, anything AR. I'm fascinated by it and I can see a ton of possibilities for it, I just can't afford to invest the time and money.
Can be used for announcements, notifications, 2-way comms etc.
If I had access to a space launch capability for less than $100/kg, however, I could bootstrap the whole business from my basement with an investment of ~$100-200k from some friends. [0]
From SpaceX's perspective, they can put ~5 tons in lunar orbit for a cost of $200k in fuel [1]. $40/kg is great, but it's not what they charge their customers, and it's not what they'd charge me - demand from comsat operators with a much higher willingness to pay means that I have no access to the market, and SpaceX needs to make a profit.
Give it 10-50 years, and I'd expect real asteroid mining will become economically viable. [2]
[0] assuming zero NRE for mining robot fabrication and a single person salary.
[1] ignoring actual launch costs, personnel salaries, and assuming the whole booster stack is paid for through infinite perfect recovery.
[2] see also Planetary Resources, and why they should be in hibernation mode, not burning $2m/year developing camera sats. but hey, VCs have money to burn.
It's actually interesting question, what price of space launch (per kg, when you have low or high kg) would enable entrepreneurs to do all sorts of crazy shit there
The idea is that people get a "bubble" but that when people vote, hopefully they take into account both quality and how similar the post is to their own. So people get a better version of their own views, which hopefully also allows them to see similar but non-identical views.
You can see that sites like reddit which tend to have high quality posts, but many subreddis have very strong biases as well. Letting people live in their own bubble means that people can write the best quality posts they can without worrying if their bias matches the subreddit they are writing in.
What, IMO, will happen is that WebAssembly will replace JavaScript, creating a level playing field for all languages. JavaScript will no longer be the only language you can on both server and client, so we will see all-Python, all-Javascript, all-Java, all-whatever frameworks in the browser (potentially each with its own GUI library, but I think that won't happen). All-Swift could be something that Apple could provide, and chances are they are already working on it.