Key Takeaways

  • Machine learning models already operate across every stage of target validation and screening, yet clinical pipelines remain stalled without mechanistic biology models.
  • Incremental 10% optimizations trap engineering teams in local minima by forcing them to accept the constraints of legacy pipelines.
  • Sal Candido credits his time at Google for proving that aiming for an order-of-magnitude leap opens up solutions that marginal targets conceal.
  • Solving translational medicine demands tackling basic disease mechanics rather than accelerating broken experimental loops.
  • Founders can bypass dead-end iterations by applying Candido's 10x First-Principles Problem Framing to their technical roadmaps.

The Candido's 10x First-Principles Problem Framing

Most teams default to shaving seconds off steps they should delete entirely. When computational biologists try to speed up drug discovery, they usually benchmark against existing lab protocols. They build faster screens, automate pipetting, or run slightly quicker docking simulations. They hunt for a 10% bump.

Sal Candido saw this mistake during his years at Google before joining the Chan Zuckerberg Biohub. When the stated objective is curing, preventing, or managing all diseases, marginal improvements hit a wall fast. You cannot optimize your way through a missing biological principle.

Here is the exact method Candido uses to reset strategy:

  • Step 1: Reject Incrementalism: Rather than targeting an incremental 10% efficiency gain on existing workflows, explicitly set the objective to a 10x order-of-magnitude leap.
  • Step 2: Expand Solution Vision: Use the radical target to take a broader view and identify completely unconsidered solution spaces that would normally be ruled out during marginal optimization.
  • Step 3: First-Principles Reconstruction: Return to first principles to determine what data, architectural breakthroughs, and basic science are fundamentally required to make the 10x leap possible.

As Candido explained during the panel, “Sometimes it is easier to approach a problem by looking at what is it going to take to make a 10x improvement rather than a 10% improvement.” That sounds counterintuitive to founders trained on agile sprints. But Candido argues that aiming for 10% locks you into your current toolchain. “And that's not because that's necessarily an easier path. It's because it allows you to take a broader view and see some solutions that you haven't been approaching, and you would go back and look from first principles.”

Pushmeet Kohli from Google DeepMind pointed out the exact same trap in drug design pipelines. Teams celebrate small computational wins without questioning whether they are moving the clinical needle. “AI is being used today already in every part of the drug discovery process,” Kohli said. “If you are asking me the question when will we see that 10x acceleration or 100x acceleration in the timelines, then the other question is what are we accelerating?”

Accelerating a flawed assay simply produces bad answers faster. Kohli noted that tools like AlphaFold cracked structural prediction, but structural prediction alone does not solve disease mechanics. “We need to tackle some of these hard challenges of biology,” Kohli said, “and only then will we be able to get these true unlocks of acceleration that dramatically transforms how drug discovery will happen.”

When This Works (and When It Doesn't)

This framing works when you confront grand scientific bottlenecks, like disease biology or energy systems, where standard engineering iterations fail to produce clinical results. If your core bottleneck is missing scientific ground truth, tuning hyperparameters on existing data yields diminishing returns.

It fails in operational phases where execution speed determines survival. If you are tuning product conversion funnels, cash collections, or server latencies, first-principles leaps waste capital. Do not reinvent databases when you need to shave 50 milliseconds off an API endpoint. Reserve this framing for your primary technical bet.

What to Do With This

Open your product sprint board this week and identify your highest-priority performance goal. Look at the key metric your team is trying to move by 10% or 15% this quarter.

Kill that ticket. Rewrite the goal as a mandatory 10x shift. If your engineering team is trying to decrease molecule screening latency from four days to three days, order them to design a system that delivers valid targets in four hours.

Watch what happens to the discussion. The team will immediately discard their current vendor contracts, their legacy python scripts, and their current assay designs because none of them can hit that number. By enforcing the 10x constraint, you force them to list the exact biological ground truth and architectural shifts required to solve the real problem.