# Editorial Feedback: “Token Math” Opening ## 1. The Strongest Idea The most surprising, counterintuitive idea in the draft is **the inversion of normal corporate spending anxiety**: at an AI-native company, the CEO is burning **more than 3x as many tokens** as the #2 user—and instead of trying to cut the bill, he’s worried his team isn’t spending *enough*. Behind that hook is a sharp, practical thesis that unifies all three sections of the newsletter (Arielle’s ops checks, Randy’s 4.5-billion-token Blender experiment, and Marcus’s model delegation): > **At the AI frontier, low token spend can be a symptom of low ambition. You should spend aggressively to test what’s possible on real bottlenecks—and only do the “token math” afterward to audit waste and right-size your models.** Right now, the headline and deck flatten this idea into generic accounting talk (`Token Math` / `What’s the ROI on experimentation?`), while the opening paragraph buries its best concrete detail in the third sentence. --- ## 2. Headline & Deck Assessment ### Current Selection * **HED:** `Token Math` * **Alt Hed:** `How Many Tokens are Too Many?` * **DEK:** `What’s the ROI on experimentation?` * **Alt DEK:** `When expensive experimentation is worth it and when to rein your tokenmaxxing in.` ### What’s not working & why 1. **`HED: Token Math` is a topic label, not a claim.** Every’s house style calls for headlines that express the most interesting or surprising claim in concise, punchy language that works as an email subject line. “Token Math” is a two-word category label that sounds like a dry pricing breakdown. Ironically, the draft’s own conclusion (in paragraph 3) is to attack bottlenecks first and *“only then do the token math.”* Making “Token Math” the headline frames the piece around bean-counting rather than ambition at the frontier. 2. **`Alt Hed: How Many Tokens are Too Many?` frames the tension backward—and as an open question.** Open-ended question headlines are generally weaker than bold claims (especially when the implied answer is “it depends”). Worse, “How many are too many?” frames the story around *cutting back*, whereas the lede’s actual hook is that the team might be spending *too few*. *(Minor copy note: “are” should be capitalized as “Are” in title case.)* 3. **`DEK: What’s the ROI on experimentation?` is abstract and repetitive.** It reads like MBA shorthand rather than a concrete implication or detail. If paired with the Alt Hed, you get two consecutive questions (`How Many Tokens are Too Many?` / `What’s the ROI on experimentation?`), which feels indecisive and gives the reader no specific reason to click. 4. **`Alt DEK: When expensive experimentation is worth it and when to rein your tokenmaxxing in.` is much closer, with two caveats:** * **Pros:** It captures the dual thrust of the draft (bold experimentation + disciplined optimization). * **Problem 1 (Definition clash):** Paragraph 2 explicitly defines *tokenmaxxing* as *“the practice of torching tokens for sport”* and states *“The goal isn’t tokenmaxxing.”* Telling readers “when to rein your tokenmaxxing in” implies they *are* torching tokens for sport and occasionally shouldn’t. * **Problem 2 (Verbal echo):** It uses *“rein … in,”* which immediately repeats *“reining yourself in”* in the first sentence of the opening paragraph. --- ## 3. Suggested Headline & Deck Alternatives Following Every’s house style—bold, subject-line-ready claims in the hed, developed with a concrete implication or detail in the dek, leaving full qualifications for the body: **Option 1: Lead with the counterintuitive ambition claim (Strongest for email)** * **HED:** You’re Probably Not Spending Enough Tokens * **DEK:** Inside Every’s rule for AI budgets: Burn tokens to find the frontier, then audit the waste. **Option 2: Lead with the “spend first, optimize second” philosophy** * **HED:** Spend the Tokens First, Do the Math Later * **DEK:** Why Every’s CEO wants the team running bigger AI experiments—even when a single failure burns 4.5 billion tokens. **Option 3: Play off the “Token Math” concept with a claim** * **HED:** The New Token Math: Why Playing It Safe Costs More * **DEK:** How we decide when an expensive AI run is worth funding—and when to send a cheaper model to do the job. **Option 4: If keeping a question/tension headline** * **HED:** Are You Burning Enough Tokens? * **DEK:** Every’s CEO uses 3x more tokens than anyone else on the team—and thinks the rest of us need to be more ambitious. --- ## 4. Opening Paragraph Assessment ### Current Opening Paragraph > Often if you are the highest spender at your company you might think about reining yourself in. But when it comes to tokens, Every CEO Dan Shipper thinks everyone else should consider ramping up. He consumes more than three times as many tokens as the next-highest user at the company. The size of the gap has him wondering whether the rest of us are being ambitious enough with AI. ### Problems to fix 1. **Hedged, slow first sentence:** *“Often if you are the highest spender at your company you might think about reining yourself in.”* Opening with *“Often if you are … you might think about …”* stacks three hedges in 16 words (and is missing a comma after *“company”*). A hypothetical truism about corporate expenses is a flat way to start; it delays the real hook and doesn’t pull the reader in from a punchy hed/dek. 2. **The best detail is buried in Sentence 3:** The most concrete, surprising fact in the opening is that **Dan consumes more than three times as many tokens as the #2 user at Every**. Putting that in the first sentence immediately grounds the piece in real data and makes the Leaderboard screenshot right below it pay off. 3. **Double-hedged wording in Sentence 2:** *“…thinks everyone else should consider ramping up.”* You want to preserve the factual qualification—Dan isn’t issuing a blanket mandate to waste money, and he’s encouraging consideration rather than forcing spend—but *“thinks … should consider”* feels bureaucratic. Pairing *“wonders whether”* or *“thinks everyone else should consider”* cleanly with the 3x stat keeps the qualification intact without dragging down the prose. 4. **Sudden POV shift in Sentence 4 (`the rest of us`):** The paragraph starts in second person (*“if you are”*), shifts to third-person reporting (*“Every CEO Dan Shipper”*), and then suddenly drops into first-person plural (*“whether the rest of us”*) in the final sentence—before reverting to third-person reporting for the entire rest of the article. Either commit to third person (*“whether the rest of the team is being ambitious enough”*) or establish the “Inside Every” insider voice intentionally. --- ## 5. Suggested Replacement Copy for the Opening All options below preserve the exact factual qualifications from the draft (Dan consumes >3x the next-highest user; he thinks others should consider ramping up / wonders if the team is being ambitious enough; and they set up paragraph 2’s qualification that the goal isn’t “tokenmaxxing” for sport). ### Option A: Lead with the 3x stat (Recommended — punchiest hook) > Every CEO Dan Shipper consumes more than three times as many tokens as the next-highest user at the company—and instead of reining himself in, he thinks everyone else should consider ramping up. Normally, topping the company expense leaderboard is a cue to cut back. When it comes to tokens, though, the size of that gap has him wondering whether the rest of the team is being ambitious enough with AI. *Why it works:* Opens with the hardest number and the central paradox in a single brisk sentence, preserves every qualification (*“thinks everyone else should consider ramping up,” “wondering whether…”*), and resolves the jarring POV shift (*“the rest of the team”* instead of *“the rest of us”*). ### Option B: Keep the “highest spender” contrast first, but tighten the prose > Usually, being the biggest spender at your company is a cue to rein yourself in. Not when it comes to tokens. Every CEO Dan Shipper consumes more than three times as many tokens as the next-highest user on the team—and instead of cutting back, he thinks everyone else should consider ramping up. The size of the gap has him wondering whether the rest of the company is being ambitious enough with AI. *Why it works:* Preserves the original opening’s rhetorical structure (the contrast between normal spending and token spending) while cutting the sluggish *“Often if you are… you might think about…”* phrasing and giving the turn a crisp beat (*“Not when it comes to tokens.”*). ### Option C: Insider “we” voice (Best if paired with Option 3/4 dek using “we/us”) > Every CEO Dan Shipper uses more than three times as many tokens as the next-highest spender on our leaderboard. In almost any other budget category, that kind of gap would be a cue for the top spender to cut back. With tokens, Dan thinks the rest of us should consider ramping up—and wonders whether we’re being ambitious enough with AI. *Why it works:* If the newsletter wants an “Inside Every” dispatch feel, introducing *“our leaderboard”* in sentence 1 makes *“the rest of us”* and *“we’re”* at the end feel intentional rather than accidental, and connects directly to the OpenAI token leaderboard screenshot that immediately follows. --- ## 6. Quick Context & Copy Notes on the Immediately Following Lines While the rest of the draft is outside the rewrite scope, two small catches worth flagging for the edit pass: * **Paragraph 3 spacing:** There is a double space after the comma in `In a conversation with head of platform Willie Williams, the two settled on a tangible starting point...` * **Typo in the Randy section:** `he dropped the implementer, caped the orchestrator at five Sol subagents` → should be **`capped`**.