/qresearch/ - Q Research★

The first place to start is with static analysis. Assume that all media platform posts will be first analyzed statically. It is easy to do as it only requires the software to act upon attributes provided by the post itself. Here is a short list of static content attributes that software could act upon in real time or during a static analysis.

User ID of poster

IP/Range of poster

Text in title/post

Domain/URL of shared link

File name

File metadata

One or more of these attributes would typically be cross referenced with a dataset of 'censored' content and the post would then be flagged/handled accordingly per the platform. This type of analysis is extremely quick and easy to do as the software doesn't need to understand the message. This is extremely important to understand BEFORE we start beating people over the head with our dankest memes. Let's look at Twitter's purge of Q posters. I cringe whenever I follow a Q link to twitter and read the comments. First, because piling on replies to a Q link is not the task we are being asked to do. We would be better served sharing the information in hostile territory. Secondly, we are making it way too easy for the platform to ID and ban at the user level. They simply harvest users from the Q post replies and ban away. Any while they are at it, they ban direct URLs to our dank memes. https://dankmeme.com/meme/123

That URL is static (doesn't change per use) and only needs to be added to the ban list once, and all other users that attempt to share it will be prevented.

Also

We know that Twitter is scraping from the User profile text too. (heard that from PVeritas)

We also have been told that only the NSA has the tech that can read the memes.

SM platforms like Twitter and FB cannot scrape the meme because of sheer volume and speed.

Thanks for posting this topic.

>>10497007 (OP)

I'm listening…

>>10499241

OK

Lawlessness Releases = People Had To See

Direct Connect Needed

WHO gave them permission to go nuts?

Now The PERSONAL Wars

Stick in Beehive = Murder Hornets

Much Make HAY - No Bail

Call Again - 1123

>>10497754

Define Digital War

"We know that Twitter is scraping from the User profile text too." Consider how these platforms make money. They are collecting and storing every piece of data they can get their hands on so they can model, analyze and predict behavior. This works very well for both targeted advertising and shadowgate style behavior manipulation. It is their bread and butter.

"SM platforms like Twitter and FB cannot scrape the meme because of sheer volume and speed." This is partially true and I plan to dive deeper on this later. They CAN do it, they just choose not to during the static analysis that happens between the user's 'click' and the content being presented to the world. Why? Response time. If a web app takes more than a few seconds to respond it frustrates the user. The important take-away from this is they don't need to deep-analyze the meme if it is named/linked as angry-pepe.jpg and has been flagged previously. It is trivial for software to identify and act upon any part of the user's request that is plain text.

I want to move on to the confounding strategies, but first want to highlight an often overlooked data point. Let's assume that like most, we leave our SM accounts logged in. When we visit from our phone or computer, it just knows who we are. (aka cookied) We are reading an article on gateway pundit about a spicy potus tweet and want to read the comments. EVERY TIME we click a link to twitter from GP, the referrer and timestamp is added to our user's identity profile. You can see it by pasting script into your URL bar on any site after navigating a link. javascript:alert(document.referrer); Certain browsers and extensions can also block or spoof the referrer information. If we were always referred from CNN and MSNBC, what would they then try to infer from our behavior? https://browsernative.com/http-referer-control-chrome/

>>10506028

Those of us over the target know.

Don't sleep.

When THEY wipe out an account for Phone/Computer.

Which THEY forced,

We know THEM now,,, ,, ,,

We knew Then Then.

The Multi Verse is Amazing.

Leveles of War Games.

International

National

Regional

States

Local

My War Needs Are REAL

Right NOW

[Contact Acquired] = Not Yet

The Shell Game Is Over.

Drop to single paddy does the job…

All Levels In Play.

Game Tragedy.

Do real farmers bale hay of Wally out?

[PCS]

We need EDu ASAP. = Lurkers Anonymous?

>>10505745

>>10505745

>Define Digital War

as apposed to Dustification?

5843

>>10506272

In Edition To

We are of Dust.

Do All.

Boom

BOom

BOOM

USA

New Nation

>>10506237

WHO is a Trojan Horse?

Troy

ER

My Call

SOON

Trash for Pay?

Play Day On The Camels Back

7th Day on The Floor

Cedars OD Lebanon

By The River

Plane Sight

Flu Over [TD]

3rd & Goal

[Ring the Bell 2017]

Nice TT Shirt

The Shell Game Is Over

Our given task is to disseminate information and truth. In the digital battlefield, the infantry of Q patriots who proudly display and share their views play an important role. Unfortunately, this also makes them easy targets.

For those eliminated from the field their only option is to return as special operators. In the land of big tech, we are the insurgents. Let's act like it.

https://en.wikipedia.org/wiki/Confounding

"confusing …causing a spurious association"

Does this sound like something we can benefit from?

When your name is DJT or Lin Wood, plenty of human eyes and hands will happily flag/censor your information manually so automation is not necessary. For the rest of us, analysis and automation are used to maximize the utility of a single content administrator. The data they rely upon to form conclusions can only be as accurate as the inputs we provide. Minimize their utility and you will limit their reach.

Start fresh.

Spin up brand new SM accounts. You'll want a seperate email and phone number for each new SM account. Reusing these does us no good as they are the foundation of the user identity.

Stop clicking the links from conservative news into social media. Your news source referrer is a great way to discriminate red vs. blue, especially if they all come from the same side. Referrer spoofing or blocking is good.

Be patient.

Your account is of no value when suspended. Keep a couple 'clean' until you need them and others 'dirty' for the pure enjoyment of haranguing the commies.

Be a paradox.

"The test of a first-rate intelligence is the ability to hold two opposed ideas in mind at the same time and still retain the ability to function."

F. Scott Fitzgerald

Software really struggles with a paradox when it seeks a single correct outcome. What does this mean to us?

Follow Trump & Biden, AOC & Jim Jordan, RNC & DNC, Fox & MSDNC, religion & atheism

Misgender yourself

Choose a different age (like 19)

Justify it as keeping your enemies closer.

Be unique.

Paraphrase, don't direct quote in posts.

Use link proxys/url shorteners for external links. The URL is now unique and also allows us to change the link target after the preview card has been generated. Results may vary here.

Use unique file names, sizes and titles. Force them to deep analyze every media post.

Be verbose.

For privacy's sake the less info you give the better. Unless, of course, that information is totally wrong. Then give them as much as they can handle. You are intentionally trying to build a massive user identity. One user doing this makes no difference, but many users doing it causes problems especially when users are analyzed as a group. At scale, this begins to invalidate the entire dataset.

Be aloof.

Instead of making clear statements, ask abstract questions.

Try "What is the most successful socialist country?" not "The road to hell is paved by socialism."

>>10507596

55

So let's use this hateful pepe meme as our sample image in a few examples. Let's assume that the platform only cares about banning pepe memes. And let's also assume we just posted pepe to twitter from a clean user account, used a unique file (name and meta) and didn't use any popular qanon hash tags to raise any flags in the static analysis layer. So far so good.

The complicated way:

https://pjreddie.com/darknet/yolo/

Consider a system like this. Instead of video frames, pictures and memes from posts are streamed in as they are posted by users in real time. So how does it know what objects to look for? Humans must 'train' the analysis model by providing a collection of images that describe each 'object' we are trying to detect in the image being analyzed. When we scour the internet for every single pepe image ever made and use them to train our model, the software creates 'comparison shortcuts' to allow dissimilar images to be compared efficiently. The speed of this process depends on multiple variables. Large image sizes, complex shapes, high accuracy thresholds and large quantities of training images to compare all add overhead and slow this down significantly. But if we were only looking for pepe, this could do it at a rate of ~30 images/sec on a single off-the-shelf GPU. I disagree with Q about "the NSA being the only ones who can do this" because I am doing it in my basement, albeit on a much smaller scale. While this type of analysis isn't currently being used, this is important to understand because this is the future state on these platforms and you can bet they have already done plenty of experimentation in this area.

The easy way:

What color is pepe? My color picker tells me he is RGB: 72,130,1 and HSV: 58,240,62. How many pixels are in the image? 300 x 300 = 90,000. What percentage of the pixels in the whole image are pepe green? Let's just call it 70%. If we create a programmatic filter that only shows us images posted from the past 24 hours where pepe green makes up 70% of the total image would the list be small enough for a human to read, analyze and flag? I think yes. Could this type of basic filtering be done in less than 1 second so that it can be done in real time during a 'static analysis'? Definitely, no AI required.

An even easier way:

What if we didn't need to inspect every single pixel? What if we only looked in one key sampling area for pepe green? If we grab a 10 x 10 square of pixels from the dead center of this picture and test them for pepe green and refilter our list of images, do we only see pepe images? Probably not. How many of these 'samples' would we need to evaluate in each image before we can positively id our pepe image? This is what Q is showing us in yesterday's meme camouflage post. Currently, it seems they are building 'composite keys' to ID the image using 4 sample areas along the image's border but the location, size and number of those samples can and will be changed in the future. Previously censored images would simply be reanalyzed using the new sampling strategy. Again, AI/deep analysis is not required to do this.

The easiest way:

What if we converted the image to black and white before sampling it? Not greyscale, but only black and white pixels. Each colored pixel needs a byte of data to describe it, but if there are only two distinct colors, we could represent those pixels with only bits. 1=white, 0=black. This vastly reduces the amount of data required to id an image from sample areas.

What is the most important lesson to be learned from all of this? BE UNIQUE!! In every example here we rely upon datasets of known image attributes to compare against. If every image being posted is unique, the software will never be able to match it to any of the known images. Every time an admin manually flags a new meme the dataset gets bigger and slower until it ceases to provide value.