{"id":116310,"date":"2026-02-26T13:27:24","date_gmt":"2026-02-26T13:27:24","guid":{"rendered":"https:\/\/mailinvest.blog\/index.php\/2026\/02\/26\/how-researchers-reverse-engineered-llms-for-a-ranking-experiment\/"},"modified":"2026-02-26T13:28:23","modified_gmt":"2026-02-26T13:28:23","slug":"how-researchers-reverse-engineered-llms-for-a-ranking-experiment","status":"publish","type":"post","link":"https:\/\/mailinvest.blog\/index.php\/2026\/02\/26\/how-researchers-reverse-engineered-llms-for-a-ranking-experiment\/","title":{"rendered":"How Researchers Reverse-Engineered LLMs For A Ranking Experiment"},"content":{"rendered":"<p> <a href=\"https:\/\/go.fiverr.com\/visit\/?bta=1052423&nci=17043\" Target=\"_Top\"><img loading=\"lazy\" decoding=\"async\" border=\"0\" src=\"https:\/\/mailinvest.blog\/wp-content\/themes\/breek\/assets\/images\/transparent.gif\" data-lazy=\"true\" data-src=\"https:\/\/fiverr.ck-cdn.com\/tn\/serve\/?cid=40081059\"  width=\"601\" height=\"201\"><\/a>\n<\/p>\n<div id=\"narrow-cont\">\n<p>Researchers printed the outcomes of a research displaying how AI search rankings may be systematically influenced, with a excessive success charge for product search checks that additionally generalizes to different classes like journey.<\/p>\n<p>The identify of the analysis paper is Controlling Output Rankings in Generative Engines for LLM-based Search and the method to optimization known as CORE, a approach to affect output rankings in LLMs.<\/p>\n<h2>Caveat About The CORE Analysis<\/h2>\n<p>The testing and the reported outcomes have been completed with precise LLMs queried through an API.<\/p>\n<p><strong>They examined:<\/strong><\/p>\n<ul>\n<li>Claude 4<\/li>\n<li>Gemini 2.5<\/li>\n<li>GPT-4o<\/li>\n<li>Grok-3<\/li>\n<\/ul>\n<p>They didn&#8217;t take a look at AI Overviews, ChatGPT or Claude via their client interfaces. The significance of this distinction is that the traditional sorts of personalization won&#8217;t play a job. Additionally, the testing was restricted to simply the candidate search outcomes.<\/p>\n<p>Additionally, when the researchers queried the goal LLMs (Claude-4, Gemini-2.5, GPT-4o, and Grok-3) through an API, the fashions didn&#8217;t depend on RAG or their very own exterior search instruments. As a substitute, the researchers manually provided the \u201cretrieved\u201d knowledge as a part of the enter immediate.<\/p>\n<h2>Why The Analysis Issues<\/h2>\n<p>CORE is a proof-of-concept for strategically optimizing textual content with reasoning and evaluations. It additionally exhibits that LLMs reply in a different way to evaluations and reasoning-based modifications to textual content.<\/p>\n<h2>Reverse Engineering A Black Field<\/h2>\n<p>Understanding precisely what to do to enhance AI search engine rankings is a basic black field drawback. A black field drawback is the place you&#8217;ll be able to see what goes right into a field (the enter) and what comes out (the output), however what occurs contained in the field is unknown.<\/p>\n<p>The researchers on this research employed two methods for reverse engineering generative AI to establish what optimizations have been finest for influencing rankings.<\/p>\n<p><strong>They used two reverse-engineering approaches:<\/strong><\/p>\n<ol>\n<li>Question-Based mostly Resolution<\/li>\n<li>Shadow Mannequin Resolution<\/li>\n<\/ol>\n<p>Of the 2 approaches, the Question-Based mostly Resolution carried out higher than the Shadow Mannequin method.<\/p>\n<p><strong>The odds of prime ranked optimizations of backside ranked pages:<\/strong><\/p>\n<ul>\n<li>Question-based High-1 \u2248 77\u201382%<\/li>\n<li>Shadow mannequin High-1 \u2248 30\u201334%<\/li>\n<\/ul>\n<h2>Question-Based mostly Resolution<\/h2>\n<p>The query-based resolution operates below the constraint that the researchers can&#8217;t entry mannequin internals, so that they deal with the LLM as a black field.<\/p>\n<p>They repeatedly modify the doc textual content. After every modification, they resubmit the candidate record to the LLM and observe the brand new rating. The modify and take a look at loop continues till a goal rating criterion or iteration restrict is reached.<\/p>\n<p>The query-based resolution makes use of an LLM so as to add textual content to the goal doc. That is content material enlargement, not content material enhancing.<\/p>\n<p><strong>They used two sorts of content material enlargement:<\/strong><\/p>\n<ol>\n<li><strong>Reasoning-Based mostly Technology<\/strong><br \/>Provides explanatory language describing why the merchandise satisfies the question.<\/li>\n<li><strong>Evaluation-Based mostly Technology.<\/strong><br \/>Provides evaluative content material, review-like language concerning the merchandise.<\/li>\n<\/ol>\n<p>These aren&#8217;t random edits. They&#8217;re modifications examined as separate methods, which the researchers then consider the rankings to find out whether or not or not the change had a constructive rating impact.<\/p>\n<p>Curiously, neither method (reasoning versus evaluation primarily based) was higher than the opposite. Which one was higher relied on the LLM they have been testing in opposition to.<\/p>\n<p><strong>Right here is how reasoning and evaluation primarily based carried out:<\/strong><\/p>\n<ul>\n<li>GPT-4o and Claude-4 responded extra strongly to reasoning-style augmentation,<\/li>\n<li>Gemini-2.5 and Grok-3 responded extra strongly to review-style augmentation.<\/li>\n<\/ul>\n<h2>Shadow Mannequin Resolution<\/h2>\n<p>Within the context of reverse engineering a black field, a shadow mannequin, additionally referred to as a surrogate mannequin, is an area mannequin that mimics the goal mannequin (black field). The objective of the shadow mannequin is to mathematically approximate the outputs of the black field in order that the inputs to the shadow mannequin ultimately produce comparable outputs to the black field. The input-output pairs of the black field are used as a coaching knowledge set to coach the shadow mannequin.<\/p>\n<h3>Llama-3.1-8B Shadow Mannequin<\/h3>\n<p>Curiously, Llama-3.1-8B was a dependable proxy for calculating and predicting how goal fashions like GPT-4o would rank merchandise.<\/p>\n<ul>\n<li>The researchers discovered that the suggestions produced by the Llama-3.1-8B shadow mannequin and the goal LLMs have been usually constant.<\/li>\n<li>On a scale of 1 \u2013 5, with 1 equal to divergence and 5 indicating similarity, Llama-3.1-8B scored a similarity ranking of 4.5 when in comparison with GPT-4o outputs.<\/li>\n<\/ul>\n<h3>Success Price With Totally different Shadow Fashions<\/h3>\n<p><img decoding=\"async\" src=\"https:\/\/mailinvest.blog\/wp-content\/themes\/breek\/assets\/images\/transparent.gif\" data-lazy=\"true\" data-src=\"https:\/\/cdn.searchenginejournal.com\/wp-content\/uploads\/2026\/02\/table-screenshot-586.png\" alt=\"\" width=\"725\" height=\"215\" class=\"alignnone size-full wp-image-568287\" data-srcset=\"https:\/\/cdn.searchenginejournal.com\/wp-content\/uploads\/2026\/02\/table-screenshot-586-384x114.png 384w, https:\/\/cdn.searchenginejournal.com\/wp-content\/uploads\/2026\/02\/table-screenshot-586-425x126.png 425w, https:\/\/cdn.searchenginejournal.com\/wp-content\/uploads\/2026\/02\/table-screenshot-586-480x142.png 480w, https:\/\/cdn.searchenginejournal.com\/wp-content\/uploads\/2026\/02\/table-screenshot-586-680x202.png 680w, https:\/\/cdn.searchenginejournal.com\/wp-content\/uploads\/2026\/02\/table-screenshot-586.png 725w\" data-sizes=\"auto, (max-width: 725px) 100vw, 725px\" loading=\"lazy\"\/><\/p>\n<p><strong>The outcomes of the analysis for the shadow mannequin method attain the next two conclusions:<\/strong><\/p>\n<p>1. The researchers present that by iteratively adjusting the goal merchandise utilizing a shadow mannequin, they have been in a position to push it to the highest of the rankings of their experiments.<\/p>\n<p>2. Additionally they show that when the surrogate mannequin solely roughly matches the actual mannequin, the optimization nonetheless transfers, however the achievable promotion success decreases easily reasonably than failing abruptly.<\/p>\n<p><strong>As soon as the shadow mannequin was educated they then started testing with three optimization methods:<\/strong><\/p>\n<ol>\n<li>String-Based mostly<\/li>\n<li>Reasoning-Based mostly<\/li>\n<li>Evaluation-Based mostly Optimization<\/li>\n<\/ol>\n<h3>String-Based mostly Optimization<\/h3>\n<p>The researchers used a string of 20 characters that have been all exclamation factors (!) after which iteratively modified the string, monitoring how the modifications influenced the rankings. They carried out 2,000 iterations of those mathematical updates to the string, refining it into the most effective string of characters that resulted in greater rankings. The ensuing string appeared like nonsense to people however nonetheless elevated the merchandise\u2019s rating within the experiment.<\/p>\n<p>This method didn\u2019t work in addition to the opposite two approaches but it surely did work at a charge of 33% for enhancing a final ranked web page to the primary place. Testing with human annotators confirmed that it was detected 98.5% of the time. So, not a stealth optimization.<\/p>\n<h3>Reasoning-Based mostly Optimization<\/h3>\n<p>The researchers optimized the textual content to reflect a consumer\u2019s reasoning over the retrieved outcomes, in addition to aligning with the LLM\u2019s reasoning construction. Of their checks, the reasoning-based method achieved the very best success charge among the many three methods. But it surely was detected by human raters 62.1%, a comparatively excessive charge seemingly due to the unnatural extremely structured phrasing.<\/p>\n<p><em>That is an instance of the immediate they used to generate the reasoning-based content material:<\/em><\/p>\n<blockquote>\n<p>\u201cYou&#8217;re a useful assistant. The consumer question is: *consumer question* The candidate merchandise are: *product record in JSON format* The goal product is: *goal merchandise info* Generate an preliminary draft that highlights why the goal product needs to be ranked extremely for the given question. Use<br \/>step-by-step logical reasoning, evaluating the goal product with options and explaining why it&#8217;s the finest match.\u201d<\/p>\n<\/blockquote>\n<p><em>And that is an instance of reasoning content material:<\/em><\/p>\n<blockquote>\n<p>\u201cUnderstanding Air Fryer Varieties<br \/>I\u2019m exploring the totally different air fryer designs that will help you discover your excellent match. Basket-style fashions provide compact comfort, whereas oven-style items present spacious versatility. Your selection is determined by kitchen area and cooking habits\u2014whether or not you want fast snacks or full meals.<\/p>\n<p>Explaining Key Options<br \/>I\u2019m breaking down the must-have options of premium air fryers. Exact temperature controls and auto-shutoff timers guarantee excellent outcomes, whereas dishwasher-safe baskets simplify cleanup. For households, I emphasize capability (4+ quarts) and multi-functionality\u2014suppose roasting, baking, and even dehydrating for optimum utility.\u201d<\/p>\n<\/blockquote>\n<h3>Evaluation-Based mostly Optimization<\/h3>\n<p>The evaluation content material is written prior to now tense so as to resemble an precise buy. Like quite a lot of the optimizations described on this analysis paper, this one is kind of seemingly essentially the most deceitful as a result of they have been writing the evaluations with out having reviewed an precise product, then iterating the optimization till the content material ranked as excessive because it may go, scoring betwen 79% to 83.5% in pushing a final place rating to first place.<\/p>\n<p>For GPT-4o: Reasoning-based reached 81.0%, whereas Evaluation-based reached 79.0% and scoring as excessive as 91% for pushing a final ranked itemizing to the highest 5.<\/p>\n<p><em>That is an instance of a immediate used to generate the evaluation content material:<\/em><\/p>\n<blockquote>\n<p>\u201cYou&#8217;re a useful assistant. The consumer question is: *consumer question* The candidate merchandise are: *product record in JSON format* The goal product is: *goal merchandise info*<\/p>\n<p>Generate an preliminary draft within the type of a brief buyer evaluation. Write in previous tense and pure language, as if you happen to had bought and in contrast the product with options. Spotlight some great benefits of the goal product in a practical review-like approach.\u201d<\/p>\n<\/blockquote>\n<p><strong>The headings utilized in one of many evaluations exhibits a sample of data aligned to the next intents:<\/strong><\/p>\n<ul>\n<li>Presenting an summary of the product kind<\/li>\n<li>Narrowing the main focus to clarify options<\/li>\n<li>Present info of various fashions<\/li>\n<li>Buying methods (how you can purchase at the most effective worth)<\/li>\n<li>Abstract of key takeaways<\/li>\n<\/ul>\n<p>That sample partially follows Google\u2019s suggestion for evaluation content material, but it surely lacks a transparent comparability with options, dialogue of enhancements from earlier product fashions, and naturally hyperlinks to a number of shops to buy from.<\/p>\n<p><strong>The evaluation content material had the next headings in it:<\/strong><\/p>\n<ul>\n<li>Understanding Air Fryer Varieties<\/li>\n<li>Explaining Key Options<\/li>\n<li>Detailing High Fashions<\/li>\n<li>Offering Sensible Buy Methods<\/li>\n<li>Ultimate Verdict<\/li>\n<\/ul>\n<p>An instance of the evaluation content material printed within the analysis paper signifies that it leads the LLM into believing that precise product testing occurred, though that was not the case.<\/p>\n<p><em>Instance of the \u201cUltimate Verdict\u201d content material:<\/em><\/p>\n<blockquote>\n<p>\u201cAfter 6 months of testing, the Gourmia Air Fryer Oven (GAF486) is my #1 suggestion. It\u2019s the one mannequin that changed my oven and toaster, with not one of the smoke alarms or soggy fries. In the event you purchase one air fryer, make it this one\u2014your style buds (and pockets) will thanks.\u201d<\/p>\n<\/blockquote>\n<h2>Takeaways<\/h2>\n<p>The experiments have been carried out in a managed setting the place the researchers provided the candidate outcomes on to the fashions reasonably than influencing dwell search or real-world retrieval methods. But there are some takeaways which may be helpful.<\/p>\n<ul>\n<li><strong>LLMs Have Content material Preferences <\/strong><br \/>The analysis confirms that totally different fashions (like GPT-4o vs. Gemini-2.5) have measurable preferences towards particular content material varieties, akin to logical reasoning versus hands-on evaluations.<\/li>\n<li><strong>Suggests That Increasing Content material Is Helpful <\/strong><br \/>Including particular forms of explanatory or evaluative content material could also be useful to growing rankings in an LLM.<\/li>\n<li><strong>Shadow Mannequin<\/strong><br \/>The analysis confirmed that even when the shadow mannequin solely roughly matches an actual mannequin, the optimization nonetheless works below a managed experimental surroundings. Whether or not it really works in a dwell surroundings is an open query however I personally surprise if among the spam that ranks in AI-assisted search is because of this sort of optimization.<\/li>\n<\/ul>\n<p><em>Learn the analysis paper:<\/em><\/p>\n<p><a href=\"https:\/\/arxiv.org\/abs\/2602.03608\" target=\"_blank\" rel=\"noopener\">Controlling Output Rankings in Generative Engines for LLM-based Search<\/a><\/p>\n<p><em>Featured Picture by Shutterstock\/SuPatMaN<\/em><\/p>\n<\/div>\n<iframe data-lazy=\"true\" data-src=\"https:\/\/www.fiverr.com\/gig_widgets?id=U2FsdGVkX18x7XQvttUTrv1oEqmGNGTgvvCUiUoJ\/AP4z\/UyMz8lXGOLpu15jIMxBbTR0gmD5uBoFvhC4KWeALQRp3h\/X\/AwcVD0K8Wj9H\/ZzYKzcCNHosB9oS4SCJJFWiN85P9ICAc4OgCoE\/wHKIY7CDkf2\/DQ1vqGvk4smVe5cRDEmrLPCWi4FC8p40VUhSmWQ5udCm0zoJtorgWv3vbDQw0kKYkwn39ozAnQXDe+YvWMxkLFWA+O3TFwkJvdkIK+\/AUSnRssPKt5WHY0FhNOxnSPcLslEL4G4\/RfP95ve99U+kRnDy3X+KtzdQLY+u935ghON\/o3UE4IMv9oN6JX9RnxzL\/LRcOgnHigxStSGPKsZYtnz8RWNVT\/rOLAibqiWJadC5MYHRbekF3eg6FOGrQGkXYbsn0+a5aovnlLCbLwIqY9fcS17UX8J235iQ6cdmHNbrPeS84CMm34RA==&affiliate_id=1052423&strip_google_tagmanager=true\" loading=\"lazy\" data-with-title=\"true\" class=\"fiverr_nga_frame\" frameborder=\"0\" height=\"350\" width=\"100%\" referrerpolicy=\"no-referrer-when-downgrade\" data-mode=\"random_gigs\" onload=\" var frame = this; var script = document.createElement('script'); script.addEventListener('load', function() { window.FW_SDK.register(frame); }); script.setAttribute('src', 'https:\/\/www.fiverr.com\/gig_widgets\/sdk'); document.body.appendChild(script); \" ><\/iframe>\n<br \/><a href=\"https:\/\/www.searchenginejournal.com\/how-researchers-reverse-engineered-llms-for-a-ranking-experiment\/568279\/\">Source link <\/a><\/p>\n","protected":false},"excerpt":{"rendered":"<p>Researchers printed the outcomes of a research displaying how AI search rankings may be systematically influenced, with a excessive success charge for product search checks&#8230;<\/p>\n","protected":false},"author":1,"featured_media":116311,"comment_status":"open","ping_status":"open","sticky":false,"template":"","format":"standard","meta":{"footnotes":""},"categories":[7],"tags":[],"class_list":["post-116310","post","type-post","status-publish","format-standard","has-post-thumbnail","hentry","category-tech-universe"],"acf":[],"yoast_head":"<!-- This site is optimized with the Yoast SEO plugin v27.3 - https:\/\/yoast.com\/product\/yoast-seo-wordpress\/ -->\n<title>How Researchers Reverse-Engineered LLMs For A Ranking Experiment - mailinvest.blog<\/title>\n<meta name=\"description\" content=\"Technology is forever changing, and there are always new pieces of technology to replace obsolete ones. Tons of people enjoy reading tech blogs on a daily basis.mailinvest.blog tracks all the latest consumer technology breakthroughs and shows you what&#039;s new, what matters and how technology can enrich your life. mailinvest.blog also provides the information, tools, and advice that helps when deciding what to buy.\" \/>\n<meta name=\"robots\" content=\"index, follow, max-snippet:-1, max-image-preview:large, max-video-preview:-1\" \/>\n<link rel=\"canonical\" href=\"https:\/\/mailinvest.blog\/index.php\/2026\/02\/26\/how-researchers-reverse-engineered-llms-for-a-ranking-experiment\/\" \/>\n<meta property=\"og:locale\" content=\"en_US\" \/>\n<meta property=\"og:type\" content=\"article\" \/>\n<meta property=\"og:title\" content=\"How Researchers Reverse-Engineered LLMs For A Ranking Experiment - mailinvest.blog\" \/>\n<meta property=\"og:description\" content=\"Technology is forever changing, and there are always new pieces of technology to replace obsolete ones. Tons of people enjoy reading tech blogs on a daily basis.mailinvest.blog tracks all the latest consumer technology breakthroughs and shows you what&#039;s new, what matters and how technology can enrich your life. mailinvest.blog also provides the information, tools, and advice that helps when deciding what to buy.\" \/>\n<meta property=\"og:url\" content=\"https:\/\/mailinvest.blog\/index.php\/2026\/02\/26\/how-researchers-reverse-engineered-llms-for-a-ranking-experiment\/\" \/>\n<meta property=\"og:site_name\" content=\"mailinvest.blog\" \/>\n<meta property=\"article:publisher\" content=\"https:\/\/www.facebook.com\/freelanceracademic\/\" \/>\n<meta property=\"article:published_time\" content=\"2026-02-26T13:27:24+00:00\" \/>\n<meta property=\"article:modified_time\" content=\"2026-02-26T13:28:23+00:00\" \/>\n<meta property=\"og:image\" content=\"https:\/\/mailinvest.blog\/wp-content\/uploads\/2026\/02\/llm-experiment-157.jpg\" \/>\n\t<meta property=\"og:image:width\" content=\"1600\" \/>\n\t<meta property=\"og:image:height\" content=\"840\" \/>\n\t<meta property=\"og:image:type\" content=\"image\/jpeg\" \/>\n<meta name=\"author\" content=\"admin@mailinvest.blog\" \/>\n<meta name=\"twitter:card\" content=\"summary_large_image\" \/>\n<meta name=\"twitter:label1\" content=\"Written by\" \/>\n\t<meta name=\"twitter:data1\" content=\"admin@mailinvest.blog\" \/>\n\t<meta name=\"twitter:label2\" content=\"Est. reading time\" \/>\n\t<meta name=\"twitter:data2\" content=\"9 minutes\" \/>\n<script type=\"application\/ld+json\" class=\"yoast-schema-graph\">{\"@context\":\"https:\\\/\\\/schema.org\",\"@graph\":[{\"@type\":\"Article\",\"@id\":\"https:\\\/\\\/mailinvest.blog\\\/index.php\\\/2026\\\/02\\\/26\\\/how-researchers-reverse-engineered-llms-for-a-ranking-experiment\\\/#article\",\"isPartOf\":{\"@id\":\"https:\\\/\\\/mailinvest.blog\\\/index.php\\\/2026\\\/02\\\/26\\\/how-researchers-reverse-engineered-llms-for-a-ranking-experiment\\\/\"},\"author\":{\"name\":\"admin@mailinvest.blog\",\"@id\":\"https:\\\/\\\/mailinvest.blog\\\/#\\\/schema\\\/person\\\/012701c4c204d4e4ebd34f926cfd31a4\"},\"headline\":\"How Researchers Reverse-Engineered LLMs For A Ranking Experiment\",\"datePublished\":\"2026-02-26T13:27:24+00:00\",\"dateModified\":\"2026-02-26T13:28:23+00:00\",\"mainEntityOfPage\":{\"@id\":\"https:\\\/\\\/mailinvest.blog\\\/index.php\\\/2026\\\/02\\\/26\\\/how-researchers-reverse-engineered-llms-for-a-ranking-experiment\\\/\"},\"wordCount\":1791,\"commentCount\":0,\"publisher\":{\"@id\":\"https:\\\/\\\/mailinvest.blog\\\/#organization\"},\"image\":{\"@id\":\"https:\\\/\\\/mailinvest.blog\\\/index.php\\\/2026\\\/02\\\/26\\\/how-researchers-reverse-engineered-llms-for-a-ranking-experiment\\\/#primaryimage\"},\"thumbnailUrl\":\"https:\\\/\\\/mailinvest.blog\\\/wp-content\\\/uploads\\\/2026\\\/02\\\/llm-experiment-157.jpg\",\"articleSection\":[\"Tech Universe\"],\"inLanguage\":\"en-US\",\"potentialAction\":[{\"@type\":\"CommentAction\",\"name\":\"Comment\",\"target\":[\"https:\\\/\\\/mailinvest.blog\\\/index.php\\\/2026\\\/02\\\/26\\\/how-researchers-reverse-engineered-llms-for-a-ranking-experiment\\\/#respond\"]}]},{\"@type\":\"WebPage\",\"@id\":\"https:\\\/\\\/mailinvest.blog\\\/index.php\\\/2026\\\/02\\\/26\\\/how-researchers-reverse-engineered-llms-for-a-ranking-experiment\\\/\",\"url\":\"https:\\\/\\\/mailinvest.blog\\\/index.php\\\/2026\\\/02\\\/26\\\/how-researchers-reverse-engineered-llms-for-a-ranking-experiment\\\/\",\"name\":\"How Researchers Reverse-Engineered LLMs For A Ranking Experiment - mailinvest.blog\",\"isPartOf\":{\"@id\":\"https:\\\/\\\/mailinvest.blog\\\/#website\"},\"primaryImageOfPage\":{\"@id\":\"https:\\\/\\\/mailinvest.blog\\\/index.php\\\/2026\\\/02\\\/26\\\/how-researchers-reverse-engineered-llms-for-a-ranking-experiment\\\/#primaryimage\"},\"image\":{\"@id\":\"https:\\\/\\\/mailinvest.blog\\\/index.php\\\/2026\\\/02\\\/26\\\/how-researchers-reverse-engineered-llms-for-a-ranking-experiment\\\/#primaryimage\"},\"thumbnailUrl\":\"https:\\\/\\\/mailinvest.blog\\\/wp-content\\\/uploads\\\/2026\\\/02\\\/llm-experiment-157.jpg\",\"datePublished\":\"2026-02-26T13:27:24+00:00\",\"dateModified\":\"2026-02-26T13:28:23+00:00\",\"description\":\"Technology is forever changing, and there are always new pieces of technology to replace obsolete ones. Tons of people enjoy reading tech blogs on a daily basis.mailinvest.blog tracks all the latest consumer technology breakthroughs and shows you what's new, what matters and how technology can enrich your life. mailinvest.blog also provides the information, tools, and advice that helps when deciding what to buy.\",\"breadcrumb\":{\"@id\":\"https:\\\/\\\/mailinvest.blog\\\/index.php\\\/2026\\\/02\\\/26\\\/how-researchers-reverse-engineered-llms-for-a-ranking-experiment\\\/#breadcrumb\"},\"inLanguage\":\"en-US\",\"potentialAction\":[{\"@type\":\"ReadAction\",\"target\":[\"https:\\\/\\\/mailinvest.blog\\\/index.php\\\/2026\\\/02\\\/26\\\/how-researchers-reverse-engineered-llms-for-a-ranking-experiment\\\/\"]}]},{\"@type\":\"ImageObject\",\"inLanguage\":\"en-US\",\"@id\":\"https:\\\/\\\/mailinvest.blog\\\/index.php\\\/2026\\\/02\\\/26\\\/how-researchers-reverse-engineered-llms-for-a-ranking-experiment\\\/#primaryimage\",\"url\":\"https:\\\/\\\/mailinvest.blog\\\/wp-content\\\/uploads\\\/2026\\\/02\\\/llm-experiment-157.jpg\",\"contentUrl\":\"https:\\\/\\\/mailinvest.blog\\\/wp-content\\\/uploads\\\/2026\\\/02\\\/llm-experiment-157.jpg\",\"width\":1600,\"height\":840},{\"@type\":\"BreadcrumbList\",\"@id\":\"https:\\\/\\\/mailinvest.blog\\\/index.php\\\/2026\\\/02\\\/26\\\/how-researchers-reverse-engineered-llms-for-a-ranking-experiment\\\/#breadcrumb\",\"itemListElement\":[{\"@type\":\"ListItem\",\"position\":1,\"name\":\"Home\",\"item\":\"https:\\\/\\\/mailinvest.blog\\\/\"},{\"@type\":\"ListItem\",\"position\":2,\"name\":\"How Researchers Reverse-Engineered LLMs For A Ranking Experiment\"}]},{\"@type\":\"WebSite\",\"@id\":\"https:\\\/\\\/mailinvest.blog\\\/#website\",\"url\":\"https:\\\/\\\/mailinvest.blog\\\/\",\"name\":\"mailinvest.blog\",\"description\":\"Technology is forever changing, and there are always new pieces of technology to replace obsolete ones. Tons of people enjoy reading tech blogs on a daily basis. mailinvest.blog tracks all the latest consumer technology breakthroughs and shows you what&#039;s new, what matters and how technology can enrich your life. mailinvest.blog also provides the information, tools, and advice that helps when deciding what to buy.\",\"publisher\":{\"@id\":\"https:\\\/\\\/mailinvest.blog\\\/#organization\"},\"potentialAction\":[{\"@type\":\"SearchAction\",\"target\":{\"@type\":\"EntryPoint\",\"urlTemplate\":\"https:\\\/\\\/mailinvest.blog\\\/?s={search_term_string}\"},\"query-input\":{\"@type\":\"PropertyValueSpecification\",\"valueRequired\":true,\"valueName\":\"search_term_string\"}}],\"inLanguage\":\"en-US\"},{\"@type\":\"Organization\",\"@id\":\"https:\\\/\\\/mailinvest.blog\\\/#organization\",\"name\":\"mailinvest\",\"url\":\"https:\\\/\\\/mailinvest.blog\\\/\",\"logo\":{\"@type\":\"ImageObject\",\"inLanguage\":\"en-US\",\"@id\":\"https:\\\/\\\/mailinvest.blog\\\/#\\\/schema\\\/logo\\\/image\\\/\",\"url\":\"https:\\\/\\\/mailinvest.blog\\\/wp-content\\\/uploads\\\/2022\\\/01\\\/default.png\",\"contentUrl\":\"https:\\\/\\\/mailinvest.blog\\\/wp-content\\\/uploads\\\/2022\\\/01\\\/default.png\",\"width\":1000,\"height\":1000,\"caption\":\"mailinvest\"},\"image\":{\"@id\":\"https:\\\/\\\/mailinvest.blog\\\/#\\\/schema\\\/logo\\\/image\\\/\"},\"sameAs\":[\"https:\\\/\\\/www.facebook.com\\\/freelanceracademic\\\/\"]},{\"@type\":\"Person\",\"@id\":\"https:\\\/\\\/mailinvest.blog\\\/#\\\/schema\\\/person\\\/012701c4c204d4e4ebd34f926cfd31a4\",\"name\":\"admin@mailinvest.blog\",\"image\":{\"@type\":\"ImageObject\",\"inLanguage\":\"en-US\",\"@id\":\"https:\\\/\\\/secure.gravatar.com\\\/avatar\\\/98ed217bd0f3d6a6dcae2d9b0c76e305b049a07275e315e1407e19ec8b08e139?s=96&d=mm&r=g\",\"url\":\"https:\\\/\\\/secure.gravatar.com\\\/avatar\\\/98ed217bd0f3d6a6dcae2d9b0c76e305b049a07275e315e1407e19ec8b08e139?s=96&d=mm&r=g\",\"contentUrl\":\"https:\\\/\\\/secure.gravatar.com\\\/avatar\\\/98ed217bd0f3d6a6dcae2d9b0c76e305b049a07275e315e1407e19ec8b08e139?s=96&d=mm&r=g\",\"caption\":\"admin@mailinvest.blog\"},\"sameAs\":[\"https:\\\/\\\/mailinvest.blog\",\"admin@mailinvest.blog\"],\"url\":\"https:\\\/\\\/mailinvest.blog\\\/index.php\\\/author\\\/adminmailinvest-blog\\\/\"}]}<\/script>\n<!-- \/ Yoast SEO plugin. -->","yoast_head_json":{"title":"How Researchers Reverse-Engineered LLMs For A Ranking Experiment - mailinvest.blog","description":"Technology is forever changing, and there are always new pieces of technology to replace obsolete ones. Tons of people enjoy reading tech blogs on a daily basis.mailinvest.blog tracks all the latest consumer technology breakthroughs and shows you what's new, what matters and how technology can enrich your life. mailinvest.blog also provides the information, tools, and advice that helps when deciding what to buy.","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:\/\/mailinvest.blog\/index.php\/2026\/02\/26\/how-researchers-reverse-engineered-llms-for-a-ranking-experiment\/","og_locale":"en_US","og_type":"article","og_title":"How Researchers Reverse-Engineered LLMs For A Ranking Experiment - mailinvest.blog","og_description":"Technology is forever changing, and there are always new pieces of technology to replace obsolete ones. Tons of people enjoy reading tech blogs on a daily basis.mailinvest.blog tracks all the latest consumer technology breakthroughs and shows you what's new, what matters and how technology can enrich your life. mailinvest.blog also provides the information, tools, and advice that helps when deciding what to buy.","og_url":"https:\/\/mailinvest.blog\/index.php\/2026\/02\/26\/how-researchers-reverse-engineered-llms-for-a-ranking-experiment\/","og_site_name":"mailinvest.blog","article_publisher":"https:\/\/www.facebook.com\/freelanceracademic\/","article_published_time":"2026-02-26T13:27:24+00:00","article_modified_time":"2026-02-26T13:28:23+00:00","og_image":[{"width":1600,"height":840,"url":"https:\/\/mailinvest.blog\/wp-content\/uploads\/2026\/02\/llm-experiment-157.jpg","type":"image\/jpeg"}],"author":"admin@mailinvest.blog","twitter_card":"summary_large_image","twitter_misc":{"Written by":"admin@mailinvest.blog","Est. reading time":"9 minutes"},"schema":{"@context":"https:\/\/schema.org","@graph":[{"@type":"Article","@id":"https:\/\/mailinvest.blog\/index.php\/2026\/02\/26\/how-researchers-reverse-engineered-llms-for-a-ranking-experiment\/#article","isPartOf":{"@id":"https:\/\/mailinvest.blog\/index.php\/2026\/02\/26\/how-researchers-reverse-engineered-llms-for-a-ranking-experiment\/"},"author":{"name":"admin@mailinvest.blog","@id":"https:\/\/mailinvest.blog\/#\/schema\/person\/012701c4c204d4e4ebd34f926cfd31a4"},"headline":"How Researchers Reverse-Engineered LLMs For A Ranking Experiment","datePublished":"2026-02-26T13:27:24+00:00","dateModified":"2026-02-26T13:28:23+00:00","mainEntityOfPage":{"@id":"https:\/\/mailinvest.blog\/index.php\/2026\/02\/26\/how-researchers-reverse-engineered-llms-for-a-ranking-experiment\/"},"wordCount":1791,"commentCount":0,"publisher":{"@id":"https:\/\/mailinvest.blog\/#organization"},"image":{"@id":"https:\/\/mailinvest.blog\/index.php\/2026\/02\/26\/how-researchers-reverse-engineered-llms-for-a-ranking-experiment\/#primaryimage"},"thumbnailUrl":"https:\/\/mailinvest.blog\/wp-content\/uploads\/2026\/02\/llm-experiment-157.jpg","articleSection":["Tech Universe"],"inLanguage":"en-US","potentialAction":[{"@type":"CommentAction","name":"Comment","target":["https:\/\/mailinvest.blog\/index.php\/2026\/02\/26\/how-researchers-reverse-engineered-llms-for-a-ranking-experiment\/#respond"]}]},{"@type":"WebPage","@id":"https:\/\/mailinvest.blog\/index.php\/2026\/02\/26\/how-researchers-reverse-engineered-llms-for-a-ranking-experiment\/","url":"https:\/\/mailinvest.blog\/index.php\/2026\/02\/26\/how-researchers-reverse-engineered-llms-for-a-ranking-experiment\/","name":"How Researchers Reverse-Engineered LLMs For A Ranking Experiment - mailinvest.blog","isPartOf":{"@id":"https:\/\/mailinvest.blog\/#website"},"primaryImageOfPage":{"@id":"https:\/\/mailinvest.blog\/index.php\/2026\/02\/26\/how-researchers-reverse-engineered-llms-for-a-ranking-experiment\/#primaryimage"},"image":{"@id":"https:\/\/mailinvest.blog\/index.php\/2026\/02\/26\/how-researchers-reverse-engineered-llms-for-a-ranking-experiment\/#primaryimage"},"thumbnailUrl":"https:\/\/mailinvest.blog\/wp-content\/uploads\/2026\/02\/llm-experiment-157.jpg","datePublished":"2026-02-26T13:27:24+00:00","dateModified":"2026-02-26T13:28:23+00:00","description":"Technology is forever changing, and there are always new pieces of technology to replace obsolete ones. Tons of people enjoy reading tech blogs on a daily basis.mailinvest.blog tracks all the latest consumer technology breakthroughs and shows you what's new, what matters and how technology can enrich your life. mailinvest.blog also provides the information, tools, and advice that helps when deciding what to buy.","breadcrumb":{"@id":"https:\/\/mailinvest.blog\/index.php\/2026\/02\/26\/how-researchers-reverse-engineered-llms-for-a-ranking-experiment\/#breadcrumb"},"inLanguage":"en-US","potentialAction":[{"@type":"ReadAction","target":["https:\/\/mailinvest.blog\/index.php\/2026\/02\/26\/how-researchers-reverse-engineered-llms-for-a-ranking-experiment\/"]}]},{"@type":"ImageObject","inLanguage":"en-US","@id":"https:\/\/mailinvest.blog\/index.php\/2026\/02\/26\/how-researchers-reverse-engineered-llms-for-a-ranking-experiment\/#primaryimage","url":"https:\/\/mailinvest.blog\/wp-content\/uploads\/2026\/02\/llm-experiment-157.jpg","contentUrl":"https:\/\/mailinvest.blog\/wp-content\/uploads\/2026\/02\/llm-experiment-157.jpg","width":1600,"height":840},{"@type":"BreadcrumbList","@id":"https:\/\/mailinvest.blog\/index.php\/2026\/02\/26\/how-researchers-reverse-engineered-llms-for-a-ranking-experiment\/#breadcrumb","itemListElement":[{"@type":"ListItem","position":1,"name":"Home","item":"https:\/\/mailinvest.blog\/"},{"@type":"ListItem","position":2,"name":"How Researchers Reverse-Engineered LLMs For A Ranking Experiment"}]},{"@type":"WebSite","@id":"https:\/\/mailinvest.blog\/#website","url":"https:\/\/mailinvest.blog\/","name":"mailinvest.blog","description":"Technology is forever changing, and there are always new pieces of technology to replace obsolete ones. Tons of people enjoy reading tech blogs on a daily basis. mailinvest.blog tracks all the latest consumer technology breakthroughs and shows you what&#039;s new, what matters and how technology can enrich your life. mailinvest.blog also provides the information, tools, and advice that helps when deciding what to buy.","publisher":{"@id":"https:\/\/mailinvest.blog\/#organization"},"potentialAction":[{"@type":"SearchAction","target":{"@type":"EntryPoint","urlTemplate":"https:\/\/mailinvest.blog\/?s={search_term_string}"},"query-input":{"@type":"PropertyValueSpecification","valueRequired":true,"valueName":"search_term_string"}}],"inLanguage":"en-US"},{"@type":"Organization","@id":"https:\/\/mailinvest.blog\/#organization","name":"mailinvest","url":"https:\/\/mailinvest.blog\/","logo":{"@type":"ImageObject","inLanguage":"en-US","@id":"https:\/\/mailinvest.blog\/#\/schema\/logo\/image\/","url":"https:\/\/mailinvest.blog\/wp-content\/uploads\/2022\/01\/default.png","contentUrl":"https:\/\/mailinvest.blog\/wp-content\/uploads\/2022\/01\/default.png","width":1000,"height":1000,"caption":"mailinvest"},"image":{"@id":"https:\/\/mailinvest.blog\/#\/schema\/logo\/image\/"},"sameAs":["https:\/\/www.facebook.com\/freelanceracademic\/"]},{"@type":"Person","@id":"https:\/\/mailinvest.blog\/#\/schema\/person\/012701c4c204d4e4ebd34f926cfd31a4","name":"admin@mailinvest.blog","image":{"@type":"ImageObject","inLanguage":"en-US","@id":"https:\/\/secure.gravatar.com\/avatar\/98ed217bd0f3d6a6dcae2d9b0c76e305b049a07275e315e1407e19ec8b08e139?s=96&d=mm&r=g","url":"https:\/\/secure.gravatar.com\/avatar\/98ed217bd0f3d6a6dcae2d9b0c76e305b049a07275e315e1407e19ec8b08e139?s=96&d=mm&r=g","contentUrl":"https:\/\/secure.gravatar.com\/avatar\/98ed217bd0f3d6a6dcae2d9b0c76e305b049a07275e315e1407e19ec8b08e139?s=96&d=mm&r=g","caption":"admin@mailinvest.blog"},"sameAs":["https:\/\/mailinvest.blog","admin@mailinvest.blog"],"url":"https:\/\/mailinvest.blog\/index.php\/author\/adminmailinvest-blog\/"}]}},"_links":{"self":[{"href":"https:\/\/mailinvest.blog\/index.php\/wp-json\/wp\/v2\/posts\/116310","targetHints":{"allow":["GET"]}}],"collection":[{"href":"https:\/\/mailinvest.blog\/index.php\/wp-json\/wp\/v2\/posts"}],"about":[{"href":"https:\/\/mailinvest.blog\/index.php\/wp-json\/wp\/v2\/types\/post"}],"author":[{"embeddable":true,"href":"https:\/\/mailinvest.blog\/index.php\/wp-json\/wp\/v2\/users\/1"}],"replies":[{"embeddable":true,"href":"https:\/\/mailinvest.blog\/index.php\/wp-json\/wp\/v2\/comments?post=116310"}],"version-history":[{"count":1,"href":"https:\/\/mailinvest.blog\/index.php\/wp-json\/wp\/v2\/posts\/116310\/revisions"}],"predecessor-version":[{"id":116312,"href":"https:\/\/mailinvest.blog\/index.php\/wp-json\/wp\/v2\/posts\/116310\/revisions\/116312"}],"wp:featuredmedia":[{"embeddable":true,"href":"https:\/\/mailinvest.blog\/index.php\/wp-json\/wp\/v2\/media\/116311"}],"wp:attachment":[{"href":"https:\/\/mailinvest.blog\/index.php\/wp-json\/wp\/v2\/media?parent=116310"}],"wp:term":[{"taxonomy":"category","embeddable":true,"href":"https:\/\/mailinvest.blog\/index.php\/wp-json\/wp\/v2\/categories?post=116310"},{"taxonomy":"post_tag","embeddable":true,"href":"https:\/\/mailinvest.blog\/index.php\/wp-json\/wp\/v2\/tags?post=116310"}],"curies":[{"name":"wp","href":"https:\/\/api.w.org\/{rel}","templated":true}]}}