Imagine you make a comic strip from just 4 images found on the internet. It doesn't contain any text yet manages to convey some meaningful narrative. Most likely, a human will be able to interpret the concept. More importantly, if you muddled up the images in a different order, a human would be able to easily identify which set of images makes more logical sense. AI is left in the dust when it comes to this challenge, and hence, this is an elegant and efficient way of proving personhood. Makes sense?!
This sort of challenge is much more "AI-Hard" when compared to standard capthcas, which AI is getting better at solving as each day passes.
Above you can see one of the classic examples of a flip on Idena. Which assembly of images do you think forms a logical story? Those on the left, or those on the right? If you guessed "left" then you are in agreement with the vast majority of people. This level of consensus can only be reached by humans.
... tells a story!
Many of the best flips follow a similar paradigm: a clear narrative with a logical flow. The four images are assembled to chronologically represent an event where something changes, and the causes and effects are obvious.
This could be described as:
We might, for example, have a boy riding his bike, followed by boy and bicycle in mid air, followed by a boy on the ground clutching a hurt knee, and finishing with an ambulance arriving at the scene, or perhaps a bandaged leg.
There is a misconception among Idena newbies that flips should be complex and challenging. The truth is rather the opposite. Flips should be easily interpreted by people. This makes for a more pleasant user experience, leads to good consensus, and still proves extremely challenging for AI.
When making your flips, focus on making them easy for humans, not hard for machines.
If your flip narrative is about a man going fishing, we don't need to see the mans pet dog, loving grandson, or saloon car.
In fact these superfluous elements can distract from the intended concept, especially with these thumbnail sized images. Remove visual distractions and let the story be obvious!
Trim the fat, and keep it simple!
You might be tempted to make a flip with 2 groups of 2, for example, "man gets money, man is happy, man loses money, man is sad", but these types of flips can be confusing when shuffled as they lack a clear beginning and end.
Likewise, be careful of "cycles" such as charging batteries, or bedtime routines. A person in bed with their eyes closed may be about to wake, or have just fallen asleep. If they're brushing their teeth in another image, that doesn't help matters much either!
This is a badly conceived flip!
So, avoid cyclical "chicken and egg" stuff. Rather let's have linear flow: the egg cracked into a frying pan, served on a plate, seasoned with pepper, and eaten!
Other examples a "linear flow" are: growing, rotting, burning and gravity.
If we see a horse rider in mid-air, we can assume they are falling from the horse, and not falling from the ground onto the horse!
That means (within reason) anyone can solve it, irrespective of their cultural background. A flip shouldn't rely on a niche cultural theme or academic concept, such as a Jewish "Bar Mitzvah", or the process of Quantum-Mechanical Tunneling! This is not about political correctness, but rather an important factor of reaching consensus as the network grows.
Text is another no-no. Having text in your flip not only alienates a portion of the audience, but weakens resistance to AI.
While AI can solve certain problems with great accuracy, it falls behind humans when it comes to matters of emotion, social interaction, and complex reasoning.
And that's it. A lot of this will hopefully seem like common sense, but it's surprising how many bad flips are circulating when you look through the Idena explorer. I'm hoping that in due course, the system will be tweaked to allow the cream rise to the top, and bury poor quality flips. Remember also to keep your flips relevant to the two seed words as much as possible, and shuffle them so the first and last images are not in the same positions.
And don't just take my word for it, these flips all either failed to reach consensus, and/or were reported.
In this flip above, I would err towards the one on the left, but this flip fails to have an obvious narrative. Having a consistent theme is not enough. It failed to reach strong consensus with a ratio of 4:3.
This flip, above, from Epoch 0033, is confusing. After a while you might pick the image on the right, assuming the idea is that you're getting increasingly closer to the scene. While it may be a "clever concept", it is not ideal for the purposes of a flip. It reached 5:1 consensus and was reported twice.
Finally, soldiers, soldiers, soldiers, more soldiers. OK, we get it. You like soldiers. This flip was rejected and reported.
When you make a bad flip, you risk damaging your reputation, damaging the Idena network, and you're leaving money (in the form of $DNA) on the table.
Hopefully this article has been helpful. It's important that newcomers learn how to create high quality flips. If you're new to Idena and are searching for an invite code, I would advise you to download Idena and get started making flips. You can then show them off on Telegram or Discord. Once someone sees that you've taken the time to get your head around flips, you'll get an invite code before long. Happy flipping.
If you would like to learn more about Idena, check out their website and get onto the Discord channel!
Main image from myaltcoins.info
Thanks especially to 𝕶𝖗𝖚𝖈𝖎𝖋𝖊𝖗, Endogen, Set Animals, and reb0rn (on the Idena Discord channel) for all the help. I will be refining this article in time. If you see any glaring mistakes or omissions, please get in touch.
