Quantifying Scientific Research Intensity

How elements-live measures research intensity across the periodic table — and why we built it the hard way.

Every element on the periodic table has a research history with a rhythm of its own. Lithium spent the 2000s fading, bottomed out around 2011, and has climbed steadily ever since. Helium's story is quieter and longer: a relative peak in the 1970s gave way to an uninterrupted slide through the mid-2020s, with only recent hints of a turnaround. These stories repeat 118 times over — in labs you've never heard of and niche conferences that meet a few times a year — long before a review article three years from now tells anyone it happened.

That lag is the problem elements-live exists to close. Citation counts take years to mature, and journal impact factors describe a venue, not a moment. Nothing standard tells you, today, whether research activity on a given piece of the periodic table — an element, a topic, an application — is rising, falling, or flat. What that gap needs is a rigorous measurement of how intense the market for ideas actually is: something like a CPI or GDP figure for research attention, systematic and consistent enough to benchmark what the world's brightest minds are spending their time on.

Two sides of one signal

Every element on elements-live is scored by one foundational metric — Research Intensity (RI) — but it's the two numbers derived from it, RIS and HRI, that actually show up on the site. Every number needs a comparison to mean anything, and RI alone doesn't specify one.

RI is the raw signal. To a first order, it's a citation-weighted sum over the scientific literature, parsed in the context of each element — corrected for the usual measurement issues any corpus this large runs into: incomplete coverage, uneven data quality, nothing exotic, just the corrections any sample built on messy real-world data needs. The harder problem, and the one that matters most, is timing: the goal is to surface new research as it happens, not just confirm what's already famous, which raw citation counts already do well enough. By studying the lifecycle of millions of real papers, elements-live uses an empirical model that handles both ends of that: discounting an older paper's fading relevance, and recognizing an early paper's promise before it shows up on anyone's most-cited list.

RIS (Research Intensity Score) is where RI becomes a benchmark — measured against every other element's RI, 0 to 1. It answers how an element's current research activity compares to the busiest element right now (Carbon, for the last 80 years). HRI (Historical Research Intensity) is RI measured against that same element's own all-time peak, not anyone else's — the more useful number for catching a shift. A quiet specialist element and a perennial leader can both read HRI near 1: Carbon peaked relative to itself about a decade ago, while Zinc has been gaining on its own history for decades. RIS ranks size. HRI tracks change.

None of the three is an arbitrary score bolted onto a paper count. They're one calculation, run at whatever reference date you ask for — today, each of the last 30 days, every month back to 1900, every year back to 1900. The 100 Years of Elements chart is that same calculation, run across the whole span at once.

How to use it

A live, corrected, decade-spanning research-intensity score only matters if it changes what you look at. A few ways it's been used so far:

• Watch HRI, not just RIS, for the real shift. RIS ranks which element is largest today — usually the ones you'd guess. HRI catches something different: a specific, maybe- unglamorous element doing what it's never done before in its own recorded history. That's the earlier signal, and the more specific one.
• Read the literature through its own structure, not a keyword search. Every element's papers are grouped into citation clusters — groups of papers that consistently cite each other, the sub-field as the literature itself defines it, rather than whatever a relevance-ranked keyword search happens to surface. That structure is also why elements-live builds research packets: a cluster turned into a briefing you can read in one sitting, or hand straight to whatever LLM you're already using, instead of rebuilding the same reading list from scratch every time.
• Check the paper count before you size anything up. RIS and HRI are both relative — to every other element, or to an element's own history — so a striking move on either can still sit on a genuinely small literature. Every element page shows exactly how many papers back its score, worth a glance before reading "rising" as "large."

Research packets are the clearest statement of what elements-live is trying to be: a deterministic layer, not another model guessing at what's relevant — a harness that turns the raw sprawl of global research activity into surfaces that show where science is moving, and why. The scores are one surface, the clusters and reading lists are another, and a packet is the third: the point where that structure crosses over into something you can hand straight to whatever AI is doing your next step, instead of asking it to guess.

Where we've drawn the line

This isn't a prediction. It doesn't say an element will matter more next year — only that its measured research intensity sits where it sits right now, with a fully-audited trail for how it got there. It has real soft spots: history before the 1990s runs at coarser time resolution than the last few years, because the underlying publication records were never collected any finer. The rarest, thinnest-literature elements — the superheavies at the bottom of the table, mostly — are noisier month to month than a well-studied element like carbon, for the plain reason that a few hundred papers swing around more than a few hundred thousand do.

Where to start

If you're going to poke at this yourself, here's the fastest path in, depending on what you're trying to answer:

• Spot a shift. Sort by HRI instead of RIS on the live table — the elements worth a second look are usually the quiet ones sitting at their own all-time high right now, not the ones already leading the pack.
• Benchmark scale. Check RIS against the obvious heavyweights — carbon, hydrogen, silicon — to see where a material you care about really sits, not where you'd assume.
• Skip the manual literature search. Pull a research packet on an element — the citation clusters and core papers, organized and ready to hand to an LLM. Zinc is a good one to start with.
• Go look up an element you work on and see if the number agrees with what you already know.

Ready to see it live? Open the table, browse any of the 118 elements, or read how to read RIS, HRI, and the rest of the numbers.