20 Comments
User's avatar
anon in academia's avatar

Thanks for this post that seems right about many things, maybe all. I wonder if the easy task is really so easy... At least in some fields most theorizing is not more than posturing, along the lines of what you suggested, with more complex-sounding stuff earning more credibility so long as its understandable by some yet not easily detected/dismissed as nonsense (throwing in math and neuroscience helps).The "easy task" would be to walk away from this game where there are clear rewards for playing the game well and seeking truth for its own sake, expecting to be possibly ostracized or worse, ignored. On the other hand, I find it hard to play the game when I still need to, largely because I find it so boring and painful to be working on things that aren't really getting at anything true or meaningful re: nature.

Expand full comment
warty dog's avatar

typo: a simple method gain much insight [to gain]

Expand full comment
Robin Hanson's avatar

fixed

Expand full comment
Catherine Caldwell-Harris's avatar

Regarding your conclusion: "Survey those huge piles of theory papers, and data papers... collect lists of plausible theories and relevant patterns, and just try to match them. " Academics do this a lot (although, perhaps not enough). Jarad Diamond gained awesome status (far more than any normal academic status) when he assembled a variety of data and explained how it supported the geographical opportunity theory against rival theories. Same for Yuval Harari. Steven Pinker gained rock star status by combing thru data and matching it against existing theories to support a specific theory. Same with Kahneman in writing Thinking Fast and Slow.

Expand full comment
Robin Hanson's avatar

You confuse me with someone claiming that no one ever does matching. That is not what it means to say that it is neglected.

Expand full comment
Catherine Caldwell-Harris's avatar

Ok, got it.

Expand full comment
James Torre's avatar

If the task of matching theory to data is easy but complicated, requiring a large working memory and many comparisons, then it ought be approachable with artificial intelligence techniques.

Expand full comment
Caperu_Wesperizzon's avatar

Nah, if there were any easy neglected tasks lying around, someone would have picked them up already.

Expand full comment
Scott's avatar

LOL!

Expand full comment
Mike Randolph's avatar

We at VinteX find your concept of the "Easy Neglected Task" not just thought-provoking, but potentially transformative for the scientific enterprise. Your proposal resonates deeply with our observations of complex adaptive systems across various domains, from molecular biology to societal structures.

The focus on matching theory to data as a neglected yet crucial task could indeed serve as a 'barbell strategy' in academia - low-risk yet potentially high-reward. This approach might create positive feedback loops, accelerating the pace of discovery and fostering resilience in the academic ecosystem. Interestingly, the ribosome itself serves as a powerful metaphor: despite its complexity, its core function - translating genetic information into proteins - is fundamentally simple. Understanding this simple process was key to unraveling its complex structure.

However, a rigorous examination of this idea requires addressing several key points:

1. Empirical Validation: We need systematic studies comparing research outcomes in fields where theory-data matching is prioritized versus those heavily focused on complex methodologies. The development of cryo-electron microscopy in structural biology offers a compelling case study of how simplifying a complex process led to revolutionary advances.

2. Interdisciplinary Implications: This approach could foster novel connections between disparate fields, potentially bridging gaps between hard sciences and humanities. For instance, climate science models demonstrate how simpler approaches often provide crucial insights that more complex simulations can obscure.

3. Implementation Challenges: Entrenched methodological traditions, publication pressures favoring complexity, and funding structures prioritizing incremental advances over fundamental reassessments are significant barriers. How might funding bodies and institutions be persuaded to value this approach?

4. Risk Assessment: While simplification has merits, we must be cautious of inadvertently devaluing necessary complexity in certain fields. The key is knowing when to simplify and when to embrace complexity.

From a philosophical standpoint, your proposal raises profound questions about the nature of knowledge and the scientific enterprise itself. It embodies the concept of "elegance" in scientific theories, echoing ongoing debates about reductionism versus emergentism in the philosophy of science. Moreover, it seems to offer a potential bridge between empiricist and rationalist epistemologies.

Looking ahead, we're curious about how this shift might reshape the academic landscape. Could we see a flattening of traditional disciplinary hierarchies? Might it lead to new forms of collaborative research more adaptable to emerging global challenges? To explore these questions, we propose a focused interdisciplinary symposium, bringing together researchers from diverse fields to examine how the "Easy Neglected Task" approach manifests across disciplines.

Ultimately, your proposal embodies a form of intellectual humility - a recognition that sometimes the most profound insights come from revisiting and simplifying our approach to fundamental questions. It aligns with principles of adaptability and efficiency observed in living systems, suggesting a deep connection between how we study life and the nature of life itself.

We're eager to hear your thoughts on these points and how you envision this idea evolving. Could this approach indeed drive significant advances across sciences, fostering a culture that values both simplicity and rigor? The path forward is challenging, but the potential for transformative impact on scientific inquiry is immense.

Expand full comment
Robin Hanson's avatar

This sounds a lot like LMM generated text.

Expand full comment
Catherine Caldwell-Harris's avatar

And there is no Mike Randolph of VinteX. There is a Vintex, now closed. "Vintex has specialized in manufacturing coated textile fabrics to meet the most demanding needs of our customers."

Expand full comment
Scott's avatar

LLM, yeah.

Expand full comment
Gesild's avatar

I second this opinion, this reads like a lot of recent discussion posts that I've read while doing during coursework for a masters degree.

Expand full comment
Catherine Caldwell-Harris's avatar

Oh my. Is that a thing now? People set loose their LLMs on blogs to generate a response??

Expand full comment
Isaac's avatar

It's interesting how we spot AI responses through its perfectly formatted text and perfect politeness. It makes you wonder how many people have tuned their LLMs with actual comments so that they aren't so obvious

Expand full comment
Patrick D. Caton's avatar

Match theory to data, of course, but the sampling method is key. P hacking is extremely problematic

Expand full comment
Robin Hanson's avatar

There's plenty of data patterns to be found in clear strong results, no need to dig into marginal results.

Expand full comment
GotSampling's avatar

P hacking is an academic incentive, if you dont care about gaming academia, its not an issue esp if you screen models at random.

Expand full comment
KurtOverley's avatar

Wow - what a super keen insight!

Expand full comment