A LESSON FOR PEDANTIC MORAL BUSYBODIES - AI TO BLAME FOR THE OBAMA-APE CLIP, NOT TRUMP
Why is it OK to show one video and not the other? The Wild Wild West of leftist media all happily laughed at the prank of lunatic comedienne Kathy Griffin holding the severed "head" of Trump with dripping blood, Democrat California Governor's Office posting the "pig" on X, but their hearts bleed for the caricature of the Obamas in ape skins that was not even posted by Trump.
As usual, the Singapore liberal crowd joined in the condemnation. All highly educated Singaporean Trump-haters jump into the bandwagon of Trump-bashing without a seconds thought of finding out what it's all about. A 60 seconds video about 2020 election fraud was posted on Trump's Truth Social media account by his publicity administrator. Unbeknownst to them, many viewers reported seeing a 2 second clip of the Obamas in ape skins at the end of the video. A few hours later that video was taken down and Trump explained he didn't see the end part so he wasn't aware of the Obama clip. Trump said there's nothing to apologise for as he had nothing to do with the Obama clip.
One popular local anti-Trump commentator, a prominent lawyer, said in his Facebook post: "But I won't be surprised if someone still defends him, because, in the age of Trump, what actually matters no longer matters anymore as long as his supporters operate freely in Trump's moral universe." That is the elite mindset, such absolutism when it comes to anything Trump. His post is typical moral judgement on the man over an incident in which he (big shot lawyer) has no clue at all. And obviously he was not interested in finding out, after all, his elite mind is above the rest, surely he cannot be wrong. Anyone offering a contrarian view that doesn't measure up to his moral standards is a blind cultish follower living in Trump's moral universe.
I come here not to bury Trump, or to praise him, or defend him, but to offer a non-partisan, non-moral judgemental technical explanation on where everyone has got it wrong about that video post. All lunatic nut-head liberal Trump-haters of Singapore should humble themselves and try to learn what actually happened. Afterall, you wanna live in the AI world of Singapore so you better understand what you are getting into.
All modern social media apps want to maximise engagement, so they often append a short clip of a recommended video right after the original video ends. This is a teaser of about 1-3 seconds of another video the platform is recommending to the viewer. This short clip is known as the auto-reel. Trump, his Truth Social account administrator, or any other content creator posting a video in their own account, has no control and does not know anything about this auto-reel.
The auto-reel is the social media platform-side function. The auto-reel is not embedded in the original video. It does not exist on Trump's video itself. Technically, it exists in the platform's backend servers as a feed logic, that is, an instruction code. There is nothing of the Obama clip that is actually attached to Trump's video.
When Trump uploads his video, the platform stores it as a fixed file on its servers. When a user (viewer) scrolls to that video in their feed, the platform's AI recommendation system selects a short snippet (the auto-reel) of another video to append or overlay at the end of the playback. This snippet is pulled from other video files stored in the servers of the platform, not from Trump's uploaded video itself. The viewer's app receives a feed package that contains the main video (Trump's original video) + a feed logic, that is, instructions to play the auto-reel snippet at the end. So the auto-reel "resides" in the platform's feed instructions, not inside the original Trump video. Nothing, not the original video nor the auto-reel snippet, exists on the viewer's device until it is rendered.
"Rendering" is the process of taking data and displaying it visually or audibly for the viewer. Before rendering, the auto-reel exists only as instruction codes on the platform servers. During rendering, the user's app downloads Trump's video and automatically fetches the 1-3 second snippet of a recommended video (in this case, the Obama video) and plays it immediately following Trump's video for the user to watch on his device screen. The snippet is gone as soon as the feed moves on.
The auto-feed is not even a fixed thingy. Not all viewers will see the same snippet. This is because auto-reels are dynamically generator per viewer based on the platform's AI algorithm. In other words, it is personalised. The platform's AI usually combines three factors to pick which video to recommend for each viewer -- viewer profile, content similarity and engagement score. Viewer profile -- shows the user's interests, history, and engagement patterns. Content similarity -- the AI looks at tags, topics, keywords, or visual similarity to the original video. (In Trump's case, the algorithm could have picked a clip from another political account.) Engage/trending score -- popular, high-performing clips are more likely to be suggested.
Thus not all viewers will see the same Obama clip when they view Trump's original video.
When a viewer downloads Trump's video, the auto-reel does not follow. Because he or she is only downloading the Trump file, not the auto-reel which exist only in the platform's feed interface.
Auto-reels are platform specific. So similarly when a viewer shares that Trump video from Truth Social to X or other platforms, the auto-reel does not travel with it. Because Truth Social's auto-reel is not part of the Trump video file. Viewers that click the shared video will instead see the auto-reels enabled by the other relevant platforms, which may not necessarily be that Obama clip.
The auto-reel is a teaser, a hook to grab viewer attention. If viewer is interested to check out the recommended video, how he activates it depends on the platform. He may tap the teaser, swipe up or sideways, or click a link.
The content creator uploading the video should follow good hygiene practice by ensuring the video ends at the specified time by trimming it of unused frames at the beginning and end of the video. If the video is trimmed, the viewer sees the final frame ends cleanly, then he sees the screen jumps immediately to something that is entirely different. It feels like the app jumped to another video. The transition is sharp and obvious. A discerning viewer would think it is a systems glitch. A non-trimmed video will have a few unused frames at the end. The viewer sees the video ends, there is a slight pause of black screen, the playback area stays active, then the auto-reel plays. The transition is seamless. The viewer thinks it may be part of the same video and is confused at the entirely different context. Trimming does not prevent auto-reels, but it controls viewer perception.
Auto reels are triggered firstly by the video reaching its end-of-playback state and secondly by the feed being in auto-advance mode (mobile feeds especially). Both are external to the one posting the video. The first trigger is platform User Interface design, the second is Viewers' own device settings. There is thus nothing that content creators who post videos directly on platforms can do.
User Interface (UI) is about the layout of the app, the video player, autoplay behavior, how transitions look and feel, and what happens when a video ends. UI is designed to guide attention, not to convey intent. The psychology of UI design is to have UI deliberately avoid hard stops, keep the playback area alive, make transitions feel seamless, reduce friction so users keep watching. This exploits human habits. People assumes continuity of the video means intention. People don't analyse player boundaries, they infer meaning from sequencing. Where it comes to politics, sequencing implies endorsement, adjacent implies intent, "What comes after" is interpreted as messaging. So when a political video (voter fraud) ends, and another face or clip appears (Obama-apes), viewer assumes "The one who posted the video chose this short clip." But in reality, it is the UI that chose it. For some hot accounts like Trump's, auto-reels can cause controversies. It is not technical malice. It is UI psychology colliding with politics. A design choice meant for platform engagement, becomes allegation, a controversy, a scandal.
So basically, the auto-reel is in the nature of advertisement by the social media platforms. As advertisers, platforms would want as vast an outreach as possible. No platform would ever bother with putting an auto-reel to any videos you or I post. They go for high profile accounts that amplify visibility - celebrities, politicians, social media influencers, journalists, controversial personalities, etc whose reach are in millions. Elon Musk has 233m followers on platform X, the biggest account. Followed by Obama with 131m. Other big accounts are Justin Bieber, Katy Perry, Rihanna, Cristiano Ronaldo, and Narendra Modi. Trump used to have 90m followers on X when he was banned in 2021. Currently Trump has 8m followers on Truth Social.
For political figures like Trump, there must be zero tolerance for UI ambiguity. They should be wary of their vulnerability which this Trump video incident clearly illustrates.. There is nothing they can do if they upload their videos to the platforms directly. One way they can avoid the debacle is to post their video on their own web. Then use a link to post on the platform. This way the platform cannot append an auto-reel to the video. When a viewer on the platform clicks an external link, he leaves the platform whose apps no longer control the data feed. But platforms penalise external feeds algorithmically so the content creator gets less hits for his post. The poster trades control for reach. That is why despite the risks, politicians continue to use direct postings to reach as many viewers as possible.
Platforms are attention-retention machines. Their business model is to retain the viewers on the platform for as long as possible. External links break their business model. When that happens, the platform loses two important things. First, no advertisements are shown while the user is off-platform. Fewer impressions per session means lower monetisation per user. Means less ad revenue. Secondly, platform can't track user behavioral data. Loss of data weakens recommendation models.
To sum up, auto-reels are feed-level features of the platform, added dynamically by AI for other users in their feeds. The uploader (Trump and his team) never see the Obama snippet when reviewing the video internally on the account where the video was uploaded. Auto-reel triggers automatically after the video ends. Seamless playback makes it look like part of the original video. High-profile visibility means many people notice the Obama snippet. Human assumptions and media amplification turn a 2 second teaser into an international scandal. They are now calling him out as a racist!
Intelligent people with Trump Dysfunction Syndrome get triggered and jump into the fray to spread nonsense against the most maligned man in human history. And when someone like me tries to provide clarity and bring sanity, it's easier to throw me into their dustbin of blinded cult followers than to do what they lambasted Trump for not doing ....... APOLOGISE.
Here is the video which the auto-reel recommended. It is an AI-generated political satire based on the animals in the cartoon film "The Lion Sleeps Tonight". It was first posted on X in October 2023 and accredited to a user known as "Xerias". Note that all Trump-haters say it is "racist" and "target the Obamas" without even having watched the video. There is nothing racist -- Biden was also an "ape". Many other politicians are portrayed as the various animals. How could apes be bad? They are the most intelligent animal after humans. Perhaps the Obamas should be portrayed as white polar bears to avoid the racist tag. Some may find the depiction distasteful.
Political cartoons are a long-standing, widely accepted part of US political culture. In US, political satire and cartoons fall under strong First Amendment (free speech) protection. The courts have repeatedly upheld the right to mock, exaggerate, insult, and caricature public figures, especially politicians. The US cultural assumption is: If you seek power, you accept ridicule. Politicians are par for the course.
Having said that, I am aware the political cartoonists of the past are cultured and capable of depicting ridicule in the most entertaining way. Today's lot tends to be crude and straying into the gutter of obscene banalities. On that score, The Lion Sleeps Tonight is a nothingburger.
This platform has withdrawn it's subscriber widget. If you like blogs like this and wish to know whenever there is a new post, click the button to my FB and follow me there. I usually intro my new blogs there. Thanks.