Making Manual Top 10

Why People Still Make Manual Top 10s When Algorithms Exist

Manual top 10 lists are still being produced at publication houses, indie record labels, retail buyers rooms, and award committees everywhere. Not because they are trendy. Because automated systems keep producing results that look fine on paper but fall apart under scrutiny. I have spent more years than I care to count watching people try to automate something that actually benefits from being slow. The core problem with algorithmic rankings is that they optimize for the wrong variables. Engagement metrics reward controversy. Revenue models reward volume. Neither of those things is what most editors or curators actually want when they commission a top 10. A manually curated list lets you weighintangibles like timing, cultural weight, personal conviction, and the kind of contextual judgment that no model trained on historical data can replicate without perpetuating past biases.

The Actual Process of Making Manual Top 10

Start with your raw pool. This is not optional. I have seen people attempt to narrow from twenty candidates directly into ten and end up with a list that reflected whoever was loudest in the room rather than whoever deserved the spots. Instead, widen first. Pull twenty-five to thirty items that meet your basic threshold criteria. If you are building a music top 10, that means everything released in your window that you actually listened to completely, not just the singles. If you are doing a retail top 10, it means every product you handled with any seriousness during the period, not just the bestsellers. Next, score or rank them on a single axis that matters for your purpose. Yes, multiple axes feel more thorough. They are not. When you introduce a second or third scoring dimension you are just hiding your real preference behind fake precision. One axis. Name it. Apply it consistently across the entire pool. I once worked on a film director top 10 where we tried to score originality, technical execution, and emotional impact separately. The list changed depending on which judge weighted originality highest. We scrapped the three-axis system entirely and just asked each person to pick what they would keep on their wall. The resulting list was faster to produce and nobody complained it was less rigorous. After you have your ranked pool, cut it to fifteen. Not ten yet. Fifteen gives you breathing room to discuss the borderline cases without the pressure of already having locked in ten spots. Now go through the fifteen-to-ten cut as a group if you are working collaboratively. If you are working alone, set the list aside for at least forty-eight hours and come back to it. The overnight review catches almost every lazy inclusion.

There is one specific edge case that catches people every single time. It happens when two items from the same artist, studio, or category compete for spots. I spent three hours once debating whether to include two tracks from the same album on a ten-list because both were genuinely strong. The rule that saved me was simple: no more than two entries from a single source unless the source has so much quality material in the pool that excluding three of its items would be indefensible. That clause did not trigger. The final list had one track from that album. It was the harder call but the correct one.

Common Mistakes That Ruin Manual Lists

The biggest mistake is recency bias. You remember the last thing you experienced more vividly, and it gets inflated in your ranking. This is not a character flaw. It is a neurological fact. The workaround is writing down your initial rankings in order before you read or hear anything else in the pool again. Lock that first pass. Then re-score from scratch after consuming the full pool. Average the two rankings. Items that stay high across both passes are genuine hits. Items that drop significantly were riding novelty. The second mistake is anchoring. Someone states their top three early in a group discussion and the rest of the group unconsciously adjusts around those choices. I have seen this destroy more good lists than any other single factor. The fix is anonymous first-round scoring. Everyone submits their top ten on paper or through a shared document without reading anyone else's. Only after all submissions are in do you compare. Even then, go around the room and have each person justify their number one pick before moving to number two. This forces actual reasoning instead of social compliance. A third mistake that people do not think about is criterion drift. Your definition of what qualifies shifts between when you start and when you finish. You begin with one standard and by the end you are applying a looser or stricter version without realizing it. Keep your criteria written down in front of you the entire time. If you change them, note the change explicitly and re-evaluate any items you already ranked. I once had a client who was ranking best-selling books of the year and midway through realized she had been implicitly filtering for literary merit when her stated goal was commercial impact. She had to restart from the qualified pool.

When Manual Top 10 Is the Wrong Call

Being honest about limitations matters. Manual top 10 lists fail when your pool exceeds roughly two hundred items and you need exhaustive coverage. At that scale the process becomes unbearably slow and the results lose consistency because human judgment fatigues. If you are ranking ten thousand products, two thousand songs, or five hundred research papers, you need a hybrid approach. Use algorithms to narrow to a manageable candidate set, then apply manual ranking on that smaller pool. Another scenario where manual fails is when the ranking axis requires domain expertise most humans do not possess. I watched a medical journal try to manually rank diagnostic tools by clinical utility and end up with wildly inconsistent results because the reviewers had different levels of clinical experience. They switched to a structured scoring rubric with explicit definitions and independent validation against patient outcomes. The results were cleaner and faster. Sometimes the right manual process is not making a raw list but building a structured decision framework. The honest takeaway is that Making Manual Top 10 works best when the domain is qualitative, the pool is under two hundred, and you actually care about nuance over speed. If you need speed or scale, use a machine. If you need judgment, do it by hand. Just do it carefully and be aware of the biases that will otherwise sneak in.

What You Actually Need to Get Started

You do not need special software. A spreadsheet, a document, or even paper works fine. What you need is a clear eligibility window, a defined output size, a single primary ranking axis, and a rule for handling ties. I usually write the tie-breaking rule before I start. It saves arguments later. Decide whether you break ties by recency, by volume, by consensus depth, or by letting the list creator make the final call. Pick one and stick to it. The list will be stronger for it. Once you have your finished top 10, publish it with context. Explain your criteria, your pool size, and any ties or near-misses you excluded. This is not extra work. It is what separates a real curated list from something that looks like a list. People trust the former and scroll past the latter. I have seen a manually ranked top 10 with detailed reasoning get more engagement and credibility than an algorithm-generated version that looked identical on the surface. The reasoning is the product. The numbers underneath are just the packaging.