Research Magnet is a research loop that complements curiosity with intelligence to attract focus to compelling research hypotheses and questions
Research has mature systems for counting outputs. It has weak systems for finding, shaping and circulating the questions that produce them.
Research has journals, grants, rankings, citations, policy papers, patents and the h-index. It has a mature machinery for counting what comes after discovery: publication counts, citation metrics, patent filings, commercialisation pathways and reckoners maintained by technology transfer and admin offices associated with most long-establishedresearch institutes. These systems are useful for the purposes they were originally created, but have long been used as proxies for values they do not represent.
Counting the countable consumes the focus of economically minded people who fund, monitor and measure R&D investments. But research has less infrastructure for what comes before the output: the good question. The pre-proposal layer is thin. Much depends on taste, institutional or data access and the serendipity of circumstances that route good questions to interested participants and back them with good resources.
Before the paper, there is a hunch. Before the patent, there is a problem. Before the grant, there is a reason to care. The real gems live in the loose ends—and the taste it takes to follow them rarely shows up in lagging research productivity metrics.
Good questions appear in the closing paragraphs of papers, where authors gesture toward the work still to be done. They sit inside rejected grant applications. They emerge from practitioners who need research to solve problems already visible in the world. They surface in hospitals, courts and classrooms.
Some sources are open. Other loops are gated. Elite labs have their own conversations. Grant panels reject proposals that others may never read. Founders, lab leaders and frontier researchers trade ideas in private networks and do not organise around the systems that guide legacy institutions. Niche communities gather in DAOs and discords that only the already-curious would know to find.
Whether we're out of ideas, or whether they're getting harder or more expensive to find is a recurring debate in metascience circles.
Not everyone loiters in the corridors of Building 20, hangs with Nat Friedman, or sits at the nerve centre of HM Treasury. Not everyone happened to catch a tweet from two undergraduate students with no formal quantum training, who swung at bitcoin with finesse.
Research Magnet is infrastructure for questions. It helps turn niche curiosity into structured inquiry: a question, a draft hypothesis, a map of the field, adjacent work, possible methods, useful datasets and facilities, relevant funders, applied contexts and potential collaborators. When complemented with intelligence, we can frame and route them to those that might make good use of them. Others can contest, endorse for fork them as they go forward, adding compute, data or parallel studies to make the research more robust.
The question might be the byproduct of operational activity, the musing of an enigma, or the frustration of an operator. When complemented with intelligence, it becomes a magnet for the attention and resources of the people and organisations that are sensitive to the signal.
Who's curious matters as much as what they're curious about. The questions a person chases and the people or institutions they bring along with them are shaped as much by their position as their personality. Many good questions need more than merit to move forward. We want to move good questions toward the people hungry for inspiration, application, or discovery. The flood of slop is the headwind; the tooling that wades through it can clarify the way.
The human–AI partnership runs both ways across the research landscape. Some fields are data-intensive—work well suited to thinking machines. Others turn on human creative judgment: the instinct that decides what's worth attention in the first place. Dependency on taste, judgment, tacit knowledge and the human ability to notice that a problem is worth caring about may be less legible to the ontologies used in the AI labs.
We can use technology to make good questions legible to good audiences. There's room for many views on how to move a question from one person's maybe-one-day list to another's daily focus. Collaborative systems can flex around values, capacity, and context. And the people who decide what gets through, supervisors, funders, stewards, operators, regulators and institutions can be supported by tools tuned to recognise promise before it becomes output.
Schumpeter claimed that the problem of capitalism is not how capital is allocated to existing structures but how structures are created and destroyed. Bankruptcy takes care of senescent businesses. But how do we get a sufficient replacement rate in systems and institutions that aren't naturally subject to extinction processes?
The relationship between the government and the community is complex. The variety of administrative legal structures suggests such an attempt might be reductionist.
The verification loop on scientific ideas is often extremely long and weirdly hostile. Ancient Athenians dismissed Aristarchus's heliocentrism in the 3rd century BC. The first successful measurement of stellar parallax was in 1838. That's a 2,000-year verification loop.
We continue to live in an answer-obsessed culture, while our career paths and life trajectories are less predictable than ever. It's easy to let uncertainty paralyze us.
Should national identity be tied to territory, culture, values—or something broader and deeper? Could it be reimagined entirely outside the structures of the state?
The Great Filter is the filter that extinguishes life. When applied to humanity, it may be that the Great Filter is in front of us in time, which means we are likely to become extinct sometime in the future. However, if the Great Filter is behind us, then we may survive extinction.
If one person is unkind to an animal it is considered to be cruelty, but where a lot of people are unkind to animals, especially in the name of commerce, the cruelty is condoned and, once large sums of money are at stake, will be defended to the last.
A philanthropic ecosystem capable of deploying $50B/year would need to be designed to identify and support high-impact opportunities, attract great talent and be transparent and accountable to the public. Design it!
Transportation policy represents an avenue for improving the systems moving people, goods, and services. Evaluating the effectiveness and benefits of existing policies on outcomes for communities may unlock progressive insights.
We do not know how breakdowns of this magnitude occur in settings with professional board members and sophisticated reporting systems. In order to prevent them, we want to understand oversight practices relating to information flow, performance oversight and risk detection.
Over the past ten years, a practice once confined to American restaurants has migrated to coffee shops around much of the world. The cause seems to have been Square's payment terminals, which made soliciting tips convenient and the default option. But was this inevitable?
A map in the rise in use of fragrance in consumer products and its correlation with childhood allergies might inform public health policy.
Schumpeter claimed that the problem of capitalism is not how capital is allocated to existing structures but how structures are created and destroyed. Bankruptcy takes care of senescent businesses. But how do we get a sufficient replacement rate in systems and institutions that aren't naturally subject to extinction processes?
The relationship between the government and the community is complex. The variety of administrative legal structures suggests such an attempt might be reductionist.
The verification loop on scientific ideas is often extremely long and weirdly hostile. Ancient Athenians dismissed Aristarchus's heliocentrism in the 3rd century BC. The first successful measurement of stellar parallax was in 1838. That's a 2,000-year verification loop.
We continue to live in an answer-obsessed culture, while our career paths and life trajectories are less predictable than ever. It's easy to let uncertainty paralyze us.
Should national identity be tied to territory, culture, values—or something broader and deeper? Could it be reimagined entirely outside the structures of the state?
The Great Filter is the filter that extinguishes life. When applied to humanity, it may be that the Great Filter is in front of us in time, which means we are likely to become extinct sometime in the future. However, if the Great Filter is behind us, then we may survive extinction.
If one person is unkind to an animal it is considered to be cruelty, but where a lot of people are unkind to animals, especially in the name of commerce, the cruelty is condoned and, once large sums of money are at stake, will be defended to the last.
A philanthropic ecosystem capable of deploying $50B/year would need to be designed to identify and support high-impact opportunities, attract great talent and be transparent and accountable to the public. Design it!
Transportation policy represents an avenue for improving the systems moving people, goods, and services. Evaluating the effectiveness and benefits of existing policies on outcomes for communities may unlock progressive insights.
We do not know how breakdowns of this magnitude occur in settings with professional board members and sophisticated reporting systems. In order to prevent them, we want to understand oversight practices relating to information flow, performance oversight and risk detection.
Over the past ten years, a practice once confined to American restaurants has migrated to coffee shops around much of the world. The cause seems to have been Square's payment terminals, which made soliciting tips convenient and the default option. But was this inevitable?
A map in the rise in use of fragrance in consumer products and its correlation with childhood allergies might inform public health policy.
Some inspiration arrives when a person with an unusual vantage point spots a gap the rest of us missed. Some comes from practitioners with real problems and a stake in seeing them solved. And some is already written down: professors keep quiet lists of theses they'd happily supervise, while the very students who'd jump at them are off trawling the literature for an idea that's sitting on that list.
The work is getting a good idea in front of a good research team at a good time and under good conditions. In private labs, confidentiality, regulatory constraints on dual-use research, and gated access to specialist data all shape what can be shared, and with whom. In the commons, the noise of the flood, the lack of context, the concealed signals of quality or interest and the increasingly decentralised structure of modern research systems dampen progress.
Some research should run on a loop. Some could be conducted by decentralised participants each contributing different components agaist a common spine. A study that once would have ended in a published paper could instead become an agent that keeps the analysis going. Building research this way takes methods and structures the old systems were never designed for.
The model takes its cues from the research subject, learning what drives a particular researcher and acting as a sparring partner—shaping ideas around that set of factors, goals and the forces pulling on their work. The structure bends to fit the research, not the other way around.
This question examines the scientific, engineering, and industrial innovations that drove major efficiency gains in gas turbines, from advanced materials and cooling to aerodynamic design and computation. It also asks how much additional efficiency improvement remains technically and economically feasible.
Gas turbines remain central to electricity generation, aviation, and industrial energy systems, so even marginal efficiency gains have large implications for fuel use, emissions, energy security, and operating cost. A clearer account of the innovation pathway can inform R&D prioritisation, technology policy, and realistic expectations about future performance gains. It also helps distinguish which advances were transformative, cumulative, or now approaching physical and economic limits.
Existing literature documents many of the component-level advances underlying turbine performance, including higher turbine inlet temperatures, superalloys, thermal barrier coatings, blade cooling, and computational design methods. However, these accounts are often technically fragmented, sector-specific, or focused on single firms, components, or periods rather than providing an integrated historical attribution of efficiency improvements across the full innovation system. There is also limited synthesis linking past innovation trajectories to prospective limits imposed by thermodynamics, materials durability, manufacturing complexity, cost, and decarbonisation pathways.
We like to think about how other people think about the next wave of research.
