MemEx / Meme Exchange general

>not real money

This. I'd actually pay to invest in memes.

That would bring up legal issues I think

issues can always be solved though

Host it with some bulletproof webhost.

Doesn't matter if there's legal issues if they can't touch you.

What legal issues?

Isn't it just an exchange where you can purchase fake stuff which have real world value?

look, breh, cops get shot, yes, but anti police rhetoric from BLM has been ramping up. cops getting killed today and eating glass. and you think it's a normal every day thing. #baka

Are you already compilling the list of memes? If so, could you give us some insight into it?

How fucked is that then?

I remember lurking and seeing that people were giving you shit for the idea a couple weeks back. I think it's actually a pretty fun idea, and /b/yards would eat it up. How does the program work?

There are too many legalities involved in handling real money. I don't want to deal with it and not interested in doing it anyway and hiding.

I want to launch with a small set of maybe (20?) and grow from there. Open to suggestions. I thinking '>tfw', 'it's happening', 'redpill me...', 'really makes you think', 'kneepads'. I need a more diverse set from boards I don't normally visit.

I think a few boards will have fun with it.

I have some scripts running on the server at intervals (maybe every hour). They analyse every board, check if the memes on the exchange are present, how often they appear, did a mod post it, does it lose or gain value from appearing on specific boards, how many replies it got, and a list of other factors all go into assigning it a value. People then invest by speculating which memes they think will go up in value.

There are safeguards to make spamming slow boards not profitable to discourage shitposting.

Not a huge fan of Raleway as the de facto font, but maybe I've just seen it too frequently. Otherwise it looks good, I really dig the color scheme.

Check out the fonts at users.teilar.gr/~g1951d/
They may or nay not be packaged for web use, but at least Aegyptus should support the hieroglyphs you're looking for.

What about "shill", "install gentoo" and "chad" (one for all boards and /g/ and /r9k/)

>that ID
> very nEET

Fuck I should steal this idea

Too late
meme.exchange

all three are good choices.

Looks great, you have an est on when you'd actually have it all working?

Lol I aint OP, I'm just reposting the link

hopefully soon, this would be cash

>I need a more diverse set from boards I don't normally visit.
turkroach or t*rkroach (although often people will just say "roach" so it might be hard to track)
t. Alberto Barbarosa
t. (in general)
___ can/can't into space
All /int/ memes

There's also "big guy" and "for you" which I think are universal.

I've had a change of heart about meme-listing and voting. I still believe there should be a group of organizers who oversee the general listing of memes. (The Shitposting and Expletives Commission)

In addition, there could be benefits to periodically holding "elections", wherein members are presented with a list of memes to be added to the exchange. Perhaps the SEC would curate that list, and they may have some veto ability (to prevent overly-broad/forced memes), but it would certainly enhance interactions between the exchange and its community.

And that's how you get from a simple free trade zone to the bureaucratic nightmare that is the EU

The easiest way to tell a computer to recognise a specific phrase is a sort of code called regular expressions. They can match anything, nothing, and everything in between. For an example, the regular expression

/(I|you) f[ea]ll for the .+ meme/i

instructs a computer to look for things like ">tfw you fell for the nvidia meme" or "how did it feel to fall for the SSD meme?"

The important part of this example is the rather innocuous ".+" bit. That tells the computer to match absolutely anything that's more than one symbol. It could be a single space, or the word "AMD", or the entire phrase "GNU/Linux, or as I've recently taken to calling it...".

So to make the exchange recognise memes, something like regular expressions needs to be used, because simply looking for single words in posts can be easily manipulated.

Now if meme listing is a direct democracy, the userbase would be at least slightly happier knowing that they determine what is and isn't qualified to be traded. But what profit-seeking user wouldn't suggest and vote for a "meme" that consists solely of that all-seeing ".+"? It would match each and every post, the value of that "meme" would skyrocket. According to the exchange, literally everyone posts it.

On the other hand, defaulting to curation means some people potentially biased administrators need to monitor every board evenly to detect new memes, which is obviously flawed in its own right.

What's the right mix? I can't know for certain, these particular circumstances have never occurred before, but I think that periodic open voting on a closed carefully-curated list is the best bet right now.

Correction: The initial (I|you) clause wouldn't work on the second example phrase. A more apt regex would just be

/f[ea]ll for the .+ meme/i

You didn't account for variations such as
>falling for the "meme" meme

Couldn't you do cryptocurrency then? It's not real money.

It was only supposed to be an example of how a regex could be exploited, but expanded to match your case, it would instead be

/f[ea]all(ing)? for the .+ meme/I

If anything this proves why having community interaction is beneficial.

How will fresh memes enter the database?

looks awesome bro. can't wait to get in on the action.

shorting "eth is a mess"
long on all things trump

IMO obviously

>ywn go back to 2012 and invest in Baneposting

bro if you sold stock in memes youd be fucking rich even at 1 cent each. grow some balls.

just make a cryptocurrency

Weeks, not days. It's a big project and I want to implement as much as possible and make it secure / scalable / properly tested etc.

What's the ___ meme? what board is it from

Regex expressions are fine. What I don't want though is users to have to suggest regex's. I would rather they voted or picked a meme and me or whoever can work out the implementation. Don't want to raise the barrier to entry.

Picked by me initially or selected based on votes afterwards. So they can have an IPO.

Depending on the country and the law, it can be real money and laws are currently being drafted to regulate it in others.

It won't involve real money.

>What's the ___ meme?
It's just a blank space. Usually there is a country's name there.
If you mean the whole sentence, the original is "Poland cannot into space," which I'm pretty sure originates from Polandball. It's from Krautchan and you can see it on /int/.

I'm going to list some /tv/ memes that should make the cut:
>Big guy
>4 you
>Bane
>JUST
>JUST FUCK MY ___ UP
>uno farto @3:17
>what did he mean by this?
>what's next for her career?
>buttered popcorn
>MAAAAAAAAAAAAAAAAAAARTHAAAAAA GET THE DOOOOOOOOOOOOOOR
>MOM'S GONNA FREAK
>ABSOLUTE MADMAN
>FRIDGEMODE
>HAIL SNYDER
>MAN OF MEAT
>BAT OF MURDER
>MAN OF MURDER
>STICK WOMAN
>again.
>Jenna Fischer.
>Absolute kino
>Kinocore
>Flick
>Joint
>KRAGER (a tripfag who is a nigger)
>wtf now I hate __________
>GORILLA JONES
>LOOK AT THIS PHOTOGRAPH
>BRRRRRRRRRRRRRRRRRRR BRRRRRRRRRRRRRRRRRRRRRRRRRRR (Jaons Bourne)
>Balding cavill
>cunny
>NOOOOOOOOOOOOOOOOOOOOOOOOOOOO!!! NOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOO!!
>Personal Falcon
>No Singles Policy

>she will never crush your balls
why even live

>skeleton
>that'll be $____
>Anvils
>Anvilposting
>AAAAAAAAAAAAAAAAAH TONY SAVE ME
>ANDEROOOOS
>Brendan JUST
>ALIMONY
>YOURE GONNA GO TO WAUUUUUUGH
>MMMMMMMMMM BOYS!!
>MMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMM

>ywn lick her toes
kill me right now

Will post more if I remember

_____ = anything goes here

Does the scraping script generate prices yet and, if it does, what kinds of prices are you seeing?

What about adding "fpbp/first post best post" and "/thread" to the "memes"

what if the memes are determined by the users
and your scrap bot starts looking the ones where most users put their money

I'll pick out the most recognisable of these for the initial list, others will be incorporated later. I visit /tv/ on occasion so I know the context for most of them.

Sort of. I'll post an update tonight though with some figures. I'm considering ditching mongo though and using potentially PostgreSQL you might be glad to hear.

Bit too easy to game and I don't want too many stocks to flood the site at once. I'd rather users voted on memes to be added to the exchange and then the meme had an "IPO" of sorts. There was talk of giving [s4s] some kind of special privilege to add a meme to the exchange of their choosing. Stuff like that isn't decided yet.

Have been quite busy but tomorrow and Friday I should have a lot of time to dedicate to making some more progress.

I am indeed happy to hear that Postgres could be taking over, especially for user account data.

Any news towards a trello/slack group?

I keep hearing conflicting things about storing time series data in SQL db's so still not 100% sure what I'm doing but most likely dropping mongo at least for the main DB anyway. You most likely know more about this than me what are you thoughts on it?

I've decided to go with the one you didn't like ha. I don't want to use slack as I use it for my normal work and don't want to mix them up and trello is just for tracking things to do so I have set up a Discord channel. Like slack it works through the browser too so people don't have to download anything if they don't want to.

If anyone wants to join they are welcome to:
discord.gg/nN3gf

Make it open source and I will implement image recognition using standard libraries. I will collect the training images myself. That is you want to make an image scraper. It doesn't have to permanently store any images though, just for recognition. Image recognition gives probabilistic meme similarity scores of images. Straight reposts and resizes would be 99% accurate. Edits of pepe and wojack would say something like 76% Pepe, 85% Wojack.

You make the scaper, I implement the recognizer. What do you say?

This is under consideration for phase 2(3?) at the moment although details are scarce. I've been researching image similarity algorithms. I want to check their accuracy first. Pepe images alone are extremely diverse. I can imagine, but it needs testing, that it would be very easy to trick an analyser into thinking an image was pepe because of the nature of it being a very simple illustration with large blocks of the same colour. Interested in your thoughts on the feasibility though?

The intention is to open source the whole project eventually.

Falling asleep now so will pick up further tomorrow.

MEME MAGIC IN THE MAKING FOLKS

An image similarity approach was not what I had in mind, but it's probably the easiest to implement.

I was thinking off a straight up object recognition approach. I would need to train a model on a labeled dataset of meme images and variants.

Right now, I wouldn't even consider anything more computationally-intensive than fuzzily matching color pallets, because there is a hell of a lot of content to be crawled on Veeky Forums (~800GB), and nobody can justify paying for AWS g2.8xlarge instances for a project like this

I've posted this before, I am seriously considering creating a real-money Meme exchange. We'll use ethereum to create new crypto coins and just create a mirror of a bitcoin exchange website. People will create their own tickers, buy into fads and trends, lose thousands. It will be a mad-house.

Who's with me?

Are you the guy from last thread? If so, good luck selling thin air.

Your idea is a retarded pyramid scheme but I have to admit you might make money if you make it crypto-based.

I wonder who's behind this post

literally "meme coin"

Nice we are really doing it

By all means go ahead.

I wonder what board is behind this post? Really makes you think.

True, however I was under the impression we would only scrape and compute hourly on only the most popular boards. This should bring down computational cost. Does anyone have an idea how many new images are posted every hour to a popular board like /pol/?

>the most popular boards
The intention is to scrape every board of which there are 70 including things like /trash/ and /qa/ etc.

That's a pretty data intensive task. Better make the scraper efficient too. An image similarity approach would be more efficient than a object recognition predictive model. Prediction itself wouldn't be so expensive, but feature extraction will.

I propose going with Structural Similarity Image Metric: trivial to implement since there is a fully functional python module: github.com/jterrace/pyssim

and a gorillion other implementations.

>Jenna Fischer.

Fucking kek'd

Im investing all my money into that

It's not too bad. Majority of boards are quite slow regarding number of new posts per hour. The scraper is being written in a way that stores the latest reply from each thread so that it only downloads threads that changed since the last time the scraper run (no more than page 1 on a lot of boards). I'm also potentially limiting it to the first few pages of each board under the logic of memes slowly heading for page 10 shouldn't be counting towards the value as the posts containing them are too stale.

What is your opinion on Veeky Forums? It is a pretty different board, and thus will outperform most other boards despite being objectivly worse. Depending on how this can be implemented, don`t you think it would be better to break Veeky Forums down into its invidual threads?

Veeky Forums is objectively the best board

>despite being objectivly worse
Why don't you hand yourself?

Outperform it how? Because it might turn out to be worth more than other boards but if it doesn't even go up in value then it's not a good investment. It's ability to increase in value is the only factor that would make it worth investing in. If it just stays consistent then it makes no difference what the price is.