{"id":6826,"date":"2026-08-26T09:00:43","date_gmt":"2026-08-26T07:00:43","guid":{"rendered":"https:\/\/stefanfritz.com\/?p=6826"},"modified":"2026-08-26T09:16:24","modified_gmt":"2026-08-26T07:16:24","slug":"masking-thresholds-in-ai","status":"publish","type":"post","link":"https:\/\/stefanfritz.com\/en\/masking-thresholds-in-ai\/","title":{"rendered":"How masking thresholds in AI systems make us stop seeing the world behind them"},"content":{"rendered":"\n

Why every metric decides which difference no longer counts<\/h2>\n\n\n\n

Forty years ago we began to digitize the world into zeros and ones. To make the world discrete, in the scientific sense. Now we are doing the same to meaning: with ontologies in processes, with our perception, and in the end with our thinking.<\/p>\n\n\n\n

That a thousand songs once fit on a device in your pocket was not because MP3 compressed the music. MP3 compressed the ear.<\/p>\n\n\n\n

That sounds like a play on words and is a technical statement. Breaking a piece of music into its frequencies makes the file not one byte smaller. The decomposition is only another way of writing the same quantity, and it was known decades before the first player existed. The file gets smaller through something else, and that something is not mathematics, it is a claim.<\/p>\n\n\n\n

What follows is not a warning. It is written for those who want to know what we are currently building all our AI systems on.<\/p>\n\n\n\n

A table that decides<\/h2>\n\n\n\n

The claim is called a psychoacoustic model<\/a>. It uses a property of hearing that everyone knows without ever having named it: a loud tone makes its quiet neighbors inaudible. Not quieter, gone. Hearing has, for every frequency band, a threshold below which nothing arrives, and that threshold moves with whatever is loud at the moment. The technical term is the masking threshold.<\/p>\n\n\n\n

The encoder computes it for every moment of the piece and throws away everything below it. What remains is a file a fraction of the original size that sounds, to a human ear, the same. The method works not because the steps are fine enough. It works because someone measured what the ear no longer distinguishes.<\/p>\n\n\n\n

The value of a codec is therefore not in the decomposition. It is in a model of the receiver.<\/p>\n\n\n\n

The condition for our AI world that no one mentions<\/h2>\n\n\n\n

The intuitive worry is that a continuous world can never arrive cleanly in discrete steps. That worry has been mathematically settled since 1949<\/a>, and in favor of the steps. A signal can be recovered from individual samples without error, not approximately but exactly.<\/p>\n\n\n\n

But only under one condition. The signal must contain nothing above a certain frequency, and you have to sample fast enough, at least twice as fast as the highest oscillation present. If the condition holds, discrete is exactly as good as continuous. If it is violated, you do not get a slightly less precise result, you get a wrong one.<\/p>\n\n\n\n

And in a particularly unpleasant way. The frequencies above the limit do not disappear, they reappear lower down, as tones that never existed in the original. In film you see the same effect in the wagon wheel that spins backward at full speed. The camera samples too slowly, and a forward motion becomes a backward one that is not there.<\/p>\n\n\n\n

A violated sampling theorem does not announce itself as an error. It announces itself, to stay with the acoustic example, as a tone. And a tone does not sound wrong, it sounds like music.<\/p>\n\n\n\n

People in test booths forty years ago who still decide what we hear<\/h2>\n\n\n\n

These masking thresholds were not derived. They were heard. People were played pairs of tones in sound-proofed rooms and pressed a button when they still perceived something. This went on for years, in an institute in Erlangen, and the phrase from the house still holds: no audio engineering without listening tests<\/a>. From those button presses came a table, and that table then decided what humanity would no longer hear on its recordings.<\/p>\n\n\n\n

The limits of the method were shown by an unaccompanied female voice. The team wore itself out on Suzanne Vega’s a cappella Tom’s Diner, because a bare voice has no loud neighbor behind which a fault could hide. Where nothing is loud, nothing is masked. The codec suddenly stood there without its most important trick.<\/p>\n\n\n\n

How deep this goes is shown by a second problem the developers sat on for years. The encoder looks at the signal in time windows, and a window is longer than a cymbal strike. If rounding happens inside such a window, it spreads over the whole window, including the silence before it. The result is a cymbal strike that, in the compressed signal, begins before it is struck.<\/p>\n\n\n\n

So the method had to be taught to cut time more finely at sudden events, so the future does not bleed into the past. That is not an image. That is the description of a standardization document.<\/p>\n\n\n\n

A method that decides what is no longer perceived is, however, familiar to most from an entirely different context. <\/p>\n\n\n\n

The company report is a codec<\/h2>\n\n\n\n

Every metric is a masking threshold. It sets which difference, from now on, is no longer one.<\/p>\n\n\n\n

A business report is, in this sense, a lossy codec of the company. It samples a few quantities, throws away everything below, and delivers a version that sounds, to the receiver, like the firm. That is not a weakness of the report, that is its job. Without compression it would be unusable, just as an uncompressed recording would be.<\/p>\n\n\n\n

The only question is whether anyone checked what lies below the threshold.<\/p>\n\n\n\n

At an IT service provider, utilization stood between 86 and 88 percent for twelve months. A clear, quiet number, confirmed every month, never a reason for a question. What the grid did not capture: which part of that utilization hung on two people who knew the customers personally, and whose judgment made the difference between a mandate and an order. When one of them left, the number stayed almost unchanged in the first reporting period and could not be held afterward. The grid measures hours. It has no field for the question of who delivers them.<\/p>\n\n\n\n

The case is not unusual. It is the rule. Two companies that look identical in the same diligence grid differ exactly where the grid does not sample, and that is usually the place where they really differ. Whoever looks at a firm only through its loud quantities does not hear the quiet ones. Not because they are missing, but because they are masked.<\/p>\n\n\n\n

None of this is new. It just rarely stands in the report you happen to be reading.<\/p>\n\n\n\n

Now judgement gets its threshold through AI<\/h2>\n\n\n\n

The same movement is running one floor up, and it is far less visible, because its terms sound like technology and not like decision.<\/p>\n\n\n\n

Whoever builds an AI system today that works without supervision has to fix two things before anything runs. First, when two runs count as the same run. A process that restarts after an interruption must not book twice; the specialists call it idempotence, and it is nothing other than a determination of what counts as the same. Second, what the result is measured against. The usable answer is the state at the end, not the path there, because the path looks different on every run.<\/p>\n\n\n\n

Both determinations are masking thresholds. They say which difference between two results is no longer one. That is entirely right, because without these thresholds nothing could be checked at all. Only no one notices that what is being decided here is which kind of difference will still exist in a firm at all.<\/p>\n\n\n\n

And here the parallel to music breaks, to our disadvantage. With the ear the receiver was physics. You could measure it, and it was done, with people in booths and a button. With judgment there is no fixed receiver. The threshold is not a measurement, it is a decision.<\/p>\n\n\n\n

It is currently decided mostly by those who supply the tools. That is no accusation, they have to, otherwise nothing works. It only means that the question of which difference still counts in a company is being answered from outside, while inside everyone talks about productivity.<\/p>\n\n\n\n

I have described elsewhere<\/a> how companies make themselves legible for their agents and give up more than they notice. The masking threshold is the next layer beneath. Above, it was decided what exists. Below, it is decided what counts as the same, and that is the sharper decision of the two, because it stays invisible.<\/p>\n\n\n\n

A recording that leaves nothing out does not exist. There are only recordings where someone knew what they were leaving out, and ones where no one knows anymore.<\/p>\n","protected":false},"excerpt":{"rendered":"

Forty years ago we began to digitize the world into zeros and ones. To make the world discrete, in the scientific sense. Now we are doing the same to meaning: with ontologies in processes, with our perception, and in the end with our thinking.<\/p>\n","protected":false},"author":2,"featured_media":6831,"comment_status":"closed","ping_status":"closed","sticky":false,"template":"","format":"standard","meta":{"footnotes":""},"categories":[491,450],"tags":[519,424,426,39,439,502,523],"class_list":["post-6826","post","type-post","status-publish","format-standard","has-post-thumbnail","hentry","category-ai","category-society","tag-deutschland-en","tag-gesellschaft-en","tag-ki-en","tag-kuenstliche-intelligenz","tag-kunstliche-intelligenz","tag-kuenstliche-intelligenz-en","tag-unternehmertum-en"],"yoast_head":"\nHow masking thresholds in AI systems make us stop seeing the world behind them - Stefan Fritz Blog<\/title>\n<meta name=\"description\" content=\"Why every metric decides which difference no longer counts\" \/>\n<meta name=\"robots\" content=\"index, follow, max-snippet:-1, max-image-preview:large, max-video-preview:-1\" \/>\n<link rel=\"canonical\" href=\"https:\/\/stefanfritz.com\/en\/masking-thresholds-in-ai\/\" \/>\n<meta property=\"og:locale\" content=\"en_US\" \/>\n<meta property=\"og:type\" content=\"article\" \/>\n<meta property=\"og:title\" content=\"How masking thresholds in AI systems make us stop seeing the world behind them - Stefan Fritz Blog\" \/>\n<meta property=\"og:description\" content=\"Why every metric decides which difference no longer counts\" \/>\n<meta property=\"og:url\" content=\"https:\/\/stefanfritz.com\/en\/masking-thresholds-in-ai\/\" \/>\n<meta property=\"og:site_name\" content=\"Stefan Fritz Blog\" \/>\n<meta property=\"article:published_time\" content=\"2026-08-26T07:00:43+00:00\" \/>\n<meta property=\"article:modified_time\" content=\"2026-08-26T07:16:24+00:00\" \/>\n<meta property=\"og:image\" content=\"https:\/\/stefanfritz.com\/wp-content\/uploads\/2026\/08\/Verdeckungsschwellen-in-KI-Systemen-1024x576.png\" \/>\n\t<meta property=\"og:image:width\" content=\"1024\" \/>\n\t<meta property=\"og:image:height\" content=\"576\" \/>\n\t<meta property=\"og:image:type\" content=\"image\/png\" \/>\n<meta name=\"author\" content=\"Stefan Fritz\" \/>\n<meta name=\"twitter:card\" content=\"summary_large_image\" \/>\n<meta name=\"twitter:label1\" content=\"Written by\" \/>\n\t<meta name=\"twitter:data1\" content=\"Stefan Fritz\" \/>\n\t<meta name=\"twitter:label2\" content=\"Est. reading time\" \/>\n\t<meta name=\"twitter:data2\" content=\"8 minutes\" \/>\n<script type=\"application\/ld+json\" class=\"yoast-schema-graph\">{\"@context\":\"https:\\\/\\\/schema.org\",\"@graph\":[{\"@type\":\"Article\",\"@id\":\"https:\\\/\\\/stefanfritz.com\\\/en\\\/masking-thresholds-in-ai\\\/#article\",\"isPartOf\":{\"@id\":\"https:\\\/\\\/stefanfritz.com\\\/en\\\/masking-thresholds-in-ai\\\/\"},\"author\":{\"name\":\"Stefan Fritz\",\"@id\":\"https:\\\/\\\/stefanfritz.com\\\/en\\\/#\\\/schema\\\/person\\\/74a8a0c6a71c3db607e12bea80825a78\"},\"headline\":\"How masking thresholds in AI systems make us stop seeing the world behind them\",\"datePublished\":\"2026-08-26T07:00:43+00:00\",\"dateModified\":\"2026-08-26T07:16:24+00:00\",\"mainEntityOfPage\":{\"@id\":\"https:\\\/\\\/stefanfritz.com\\\/en\\\/masking-thresholds-in-ai\\\/\"},\"wordCount\":1408,\"publisher\":{\"@id\":\"https:\\\/\\\/stefanfritz.com\\\/en\\\/#\\\/schema\\\/person\\\/74a8a0c6a71c3db607e12bea80825a78\"},\"image\":{\"@id\":\"https:\\\/\\\/stefanfritz.com\\\/en\\\/masking-thresholds-in-ai\\\/#primaryimage\"},\"thumbnailUrl\":\"https:\\\/\\\/stefanfritz.com\\\/wp-content\\\/uploads\\\/2026\\\/08\\\/Verdeckungsschwellen-in-KI-Systemen.png\",\"keywords\":[\"Deutschland\",\"Gesellschaft\",\"KI\",\"K\u00fcnstliche Intelligenz\",\"K\u00fcnstliche Intelligenz\",\"K\u00fcnstliche Intelligenz\",\"Unternehmertum\"],\"articleSection\":[\"AI\",\"Society\"],\"inLanguage\":\"en-US\"},{\"@type\":\"WebPage\",\"@id\":\"https:\\\/\\\/stefanfritz.com\\\/en\\\/masking-thresholds-in-ai\\\/\",\"url\":\"https:\\\/\\\/stefanfritz.com\\\/en\\\/masking-thresholds-in-ai\\\/\",\"name\":\"How masking thresholds in AI systems make us stop seeing the world behind them - Stefan Fritz Blog\",\"isPartOf\":{\"@id\":\"https:\\\/\\\/stefanfritz.com\\\/en\\\/#website\"},\"primaryImageOfPage\":{\"@id\":\"https:\\\/\\\/stefanfritz.com\\\/en\\\/masking-thresholds-in-ai\\\/#primaryimage\"},\"image\":{\"@id\":\"https:\\\/\\\/stefanfritz.com\\\/en\\\/masking-thresholds-in-ai\\\/#primaryimage\"},\"thumbnailUrl\":\"https:\\\/\\\/stefanfritz.com\\\/wp-content\\\/uploads\\\/2026\\\/08\\\/Verdeckungsschwellen-in-KI-Systemen.png\",\"datePublished\":\"2026-08-26T07:00:43+00:00\",\"dateModified\":\"2026-08-26T07:16:24+00:00\",\"description\":\"Why every metric decides which difference no longer counts\",\"breadcrumb\":{\"@id\":\"https:\\\/\\\/stefanfritz.com\\\/en\\\/masking-thresholds-in-ai\\\/#breadcrumb\"},\"inLanguage\":\"en-US\",\"potentialAction\":[{\"@type\":\"ReadAction\",\"target\":[\"https:\\\/\\\/stefanfritz.com\\\/en\\\/masking-thresholds-in-ai\\\/\"]}]},{\"@type\":\"ImageObject\",\"inLanguage\":\"en-US\",\"@id\":\"https:\\\/\\\/stefanfritz.com\\\/en\\\/masking-thresholds-in-ai\\\/#primaryimage\",\"url\":\"https:\\\/\\\/stefanfritz.com\\\/wp-content\\\/uploads\\\/2026\\\/08\\\/Verdeckungsschwellen-in-KI-Systemen.png\",\"contentUrl\":\"https:\\\/\\\/stefanfritz.com\\\/wp-content\\\/uploads\\\/2026\\\/08\\\/Verdeckungsschwellen-in-KI-Systemen.png\",\"width\":1672,\"height\":941},{\"@type\":\"BreadcrumbList\",\"@id\":\"https:\\\/\\\/stefanfritz.com\\\/en\\\/masking-thresholds-in-ai\\\/#breadcrumb\",\"itemListElement\":[{\"@type\":\"ListItem\",\"position\":1,\"name\":\"Startseite\",\"item\":\"https:\\\/\\\/stefanfritz.com\\\/en\\\/\"},{\"@type\":\"ListItem\",\"position\":2,\"name\":\"How masking thresholds in AI systems make us stop seeing the world behind them\"}]},{\"@type\":\"WebSite\",\"@id\":\"https:\\\/\\\/stefanfritz.com\\\/en\\\/#website\",\"url\":\"https:\\\/\\\/stefanfritz.com\\\/en\\\/\",\"name\":\"Stefan Fritz Blog\",\"description\":\"DeepTech, Impact Investing & People\",\"publisher\":{\"@id\":\"https:\\\/\\\/stefanfritz.com\\\/en\\\/#\\\/schema\\\/person\\\/74a8a0c6a71c3db607e12bea80825a78\"},\"potentialAction\":[{\"@type\":\"SearchAction\",\"target\":{\"@type\":\"EntryPoint\",\"urlTemplate\":\"https:\\\/\\\/stefanfritz.com\\\/en\\\/?s={search_term_string}\"},\"query-input\":{\"@type\":\"PropertyValueSpecification\",\"valueRequired\":true,\"valueName\":\"search_term_string\"}}],\"inLanguage\":\"en-US\"},{\"@type\":[\"Person\",\"Organization\"],\"@id\":\"https:\\\/\\\/stefanfritz.com\\\/en\\\/#\\\/schema\\\/person\\\/74a8a0c6a71c3db607e12bea80825a78\",\"name\":\"Stefan Fritz\",\"image\":{\"@type\":\"ImageObject\",\"inLanguage\":\"en-US\",\"@id\":\"\\\/wp-content\\\/uploads\\\/2022\\\/09\\\/Stefan-Fritz-Logo-Color.png\",\"url\":\"\\\/wp-content\\\/uploads\\\/2022\\\/09\\\/Stefan-Fritz-Logo-Color.png\",\"contentUrl\":\"\\\/wp-content\\\/uploads\\\/2022\\\/09\\\/Stefan-Fritz-Logo-Color.png\",\"width\":1500,\"height\":345,\"caption\":\"Stefan Fritz\"},\"logo\":{\"@id\":\"\\\/wp-content\\\/uploads\\\/2022\\\/09\\\/Stefan-Fritz-Logo-Color.png\"}}]}<\/script>\n<!-- \/ Yoast SEO plugin. -->","yoast_head_json":{"title":"How masking thresholds in AI systems make us stop seeing the world behind them - Stefan Fritz Blog","description":"Why every metric decides which difference no longer counts","robots":{"index":"index","follow":"follow","max-snippet":"max-snippet:-1","max-image-preview":"max-image-preview:large","max-video-preview":"max-video-preview:-1"},"canonical":"https:\/\/stefanfritz.com\/en\/masking-thresholds-in-ai\/","og_locale":"en_US","og_type":"article","og_title":"How masking thresholds in AI systems make us stop seeing the world behind them - Stefan Fritz Blog","og_description":"Why every metric decides which difference no longer counts","og_url":"https:\/\/stefanfritz.com\/en\/masking-thresholds-in-ai\/","og_site_name":"Stefan Fritz Blog","article_published_time":"2026-08-26T07:00:43+00:00","article_modified_time":"2026-08-26T07:16:24+00:00","og_image":[{"width":1024,"height":576,"url":"https:\/\/stefanfritz.com\/wp-content\/uploads\/2026\/08\/Verdeckungsschwellen-in-KI-Systemen-1024x576.png","type":"image\/png"}],"author":"Stefan Fritz","twitter_card":"summary_large_image","twitter_misc":{"Written by":"Stefan Fritz","Est. reading time":"8 minutes"},"schema":{"@context":"https:\/\/schema.org","@graph":[{"@type":"Article","@id":"https:\/\/stefanfritz.com\/en\/masking-thresholds-in-ai\/#article","isPartOf":{"@id":"https:\/\/stefanfritz.com\/en\/masking-thresholds-in-ai\/"},"author":{"name":"Stefan Fritz","@id":"https:\/\/stefanfritz.com\/en\/#\/schema\/person\/74a8a0c6a71c3db607e12bea80825a78"},"headline":"How masking thresholds in AI systems make us stop seeing the world behind them","datePublished":"2026-08-26T07:00:43+00:00","dateModified":"2026-08-26T07:16:24+00:00","mainEntityOfPage":{"@id":"https:\/\/stefanfritz.com\/en\/masking-thresholds-in-ai\/"},"wordCount":1408,"publisher":{"@id":"https:\/\/stefanfritz.com\/en\/#\/schema\/person\/74a8a0c6a71c3db607e12bea80825a78"},"image":{"@id":"https:\/\/stefanfritz.com\/en\/masking-thresholds-in-ai\/#primaryimage"},"thumbnailUrl":"https:\/\/stefanfritz.com\/wp-content\/uploads\/2026\/08\/Verdeckungsschwellen-in-KI-Systemen.png","keywords":["Deutschland","Gesellschaft","KI","K\u00fcnstliche Intelligenz","K\u00fcnstliche Intelligenz","K\u00fcnstliche Intelligenz","Unternehmertum"],"articleSection":["AI","Society"],"inLanguage":"en-US"},{"@type":"WebPage","@id":"https:\/\/stefanfritz.com\/en\/masking-thresholds-in-ai\/","url":"https:\/\/stefanfritz.com\/en\/masking-thresholds-in-ai\/","name":"How masking thresholds in AI systems make us stop seeing the world behind them - Stefan Fritz Blog","isPartOf":{"@id":"https:\/\/stefanfritz.com\/en\/#website"},"primaryImageOfPage":{"@id":"https:\/\/stefanfritz.com\/en\/masking-thresholds-in-ai\/#primaryimage"},"image":{"@id":"https:\/\/stefanfritz.com\/en\/masking-thresholds-in-ai\/#primaryimage"},"thumbnailUrl":"https:\/\/stefanfritz.com\/wp-content\/uploads\/2026\/08\/Verdeckungsschwellen-in-KI-Systemen.png","datePublished":"2026-08-26T07:00:43+00:00","dateModified":"2026-08-26T07:16:24+00:00","description":"Why every metric decides which difference no longer counts","breadcrumb":{"@id":"https:\/\/stefanfritz.com\/en\/masking-thresholds-in-ai\/#breadcrumb"},"inLanguage":"en-US","potentialAction":[{"@type":"ReadAction","target":["https:\/\/stefanfritz.com\/en\/masking-thresholds-in-ai\/"]}]},{"@type":"ImageObject","inLanguage":"en-US","@id":"https:\/\/stefanfritz.com\/en\/masking-thresholds-in-ai\/#primaryimage","url":"https:\/\/stefanfritz.com\/wp-content\/uploads\/2026\/08\/Verdeckungsschwellen-in-KI-Systemen.png","contentUrl":"https:\/\/stefanfritz.com\/wp-content\/uploads\/2026\/08\/Verdeckungsschwellen-in-KI-Systemen.png","width":1672,"height":941},{"@type":"BreadcrumbList","@id":"https:\/\/stefanfritz.com\/en\/masking-thresholds-in-ai\/#breadcrumb","itemListElement":[{"@type":"ListItem","position":1,"name":"Startseite","item":"https:\/\/stefanfritz.com\/en\/"},{"@type":"ListItem","position":2,"name":"How masking thresholds in AI systems make us stop seeing the world behind them"}]},{"@type":"WebSite","@id":"https:\/\/stefanfritz.com\/en\/#website","url":"https:\/\/stefanfritz.com\/en\/","name":"Stefan Fritz Blog","description":"DeepTech, Impact Investing & People","publisher":{"@id":"https:\/\/stefanfritz.com\/en\/#\/schema\/person\/74a8a0c6a71c3db607e12bea80825a78"},"potentialAction":[{"@type":"SearchAction","target":{"@type":"EntryPoint","urlTemplate":"https:\/\/stefanfritz.com\/en\/?s={search_term_string}"},"query-input":{"@type":"PropertyValueSpecification","valueRequired":true,"valueName":"search_term_string"}}],"inLanguage":"en-US"},{"@type":["Person","Organization"],"@id":"https:\/\/stefanfritz.com\/en\/#\/schema\/person\/74a8a0c6a71c3db607e12bea80825a78","name":"Stefan Fritz","image":{"@type":"ImageObject","inLanguage":"en-US","@id":"\/wp-content\/uploads\/2022\/09\/Stefan-Fritz-Logo-Color.png","url":"\/wp-content\/uploads\/2022\/09\/Stefan-Fritz-Logo-Color.png","contentUrl":"\/wp-content\/uploads\/2022\/09\/Stefan-Fritz-Logo-Color.png","width":1500,"height":345,"caption":"Stefan Fritz"},"logo":{"@id":"\/wp-content\/uploads\/2022\/09\/Stefan-Fritz-Logo-Color.png"}}]}},"_links":{"self":[{"href":"https:\/\/stefanfritz.com\/en\/wp-json\/wp\/v2\/posts\/6826","targetHints":{"allow":["GET"]}}],"collection":[{"href":"https:\/\/stefanfritz.com\/en\/wp-json\/wp\/v2\/posts"}],"about":[{"href":"https:\/\/stefanfritz.com\/en\/wp-json\/wp\/v2\/types\/post"}],"author":[{"embeddable":true,"href":"https:\/\/stefanfritz.com\/en\/wp-json\/wp\/v2\/users\/2"}],"replies":[{"embeddable":true,"href":"https:\/\/stefanfritz.com\/en\/wp-json\/wp\/v2\/comments?post=6826"}],"version-history":[{"count":3,"href":"https:\/\/stefanfritz.com\/en\/wp-json\/wp\/v2\/posts\/6826\/revisions"}],"predecessor-version":[{"id":6832,"href":"https:\/\/stefanfritz.com\/en\/wp-json\/wp\/v2\/posts\/6826\/revisions\/6832"}],"wp:featuredmedia":[{"embeddable":true,"href":"https:\/\/stefanfritz.com\/en\/wp-json\/wp\/v2\/media\/6831"}],"wp:attachment":[{"href":"https:\/\/stefanfritz.com\/en\/wp-json\/wp\/v2\/media?parent=6826"}],"wp:term":[{"taxonomy":"category","embeddable":true,"href":"https:\/\/stefanfritz.com\/en\/wp-json\/wp\/v2\/categories?post=6826"},{"taxonomy":"post_tag","embeddable":true,"href":"https:\/\/stefanfritz.com\/en\/wp-json\/wp\/v2\/tags?post=6826"}],"curies":[{"name":"wp","href":"https:\/\/api.w.org\/{rel}","templated":true}]}}