ssivark

Nice work putting this together! But also, we're entering an era where folks might easily be able to scratch their own itch for many of these needs :-)

The striking detail here is that the patient's post-shift numbers wouldn't have tripped the static alarms we currently use; a resting heart rate moving from 44 to 49 bpm is still "athletic" by population standards. What mattered was the synchronized shift in several metrics relative to his personal baseline.

Perhaps because statistics education attunes us to average over individual differences, any mention of "data" unfortunately makes people instinctively jump to this 'cross-sectional' corner... textbooks explain how the typical human body behaves, and standard education and medical practice respectively emphasize averaging and population-level guidelines.

I think this approach is limited in some fundamental ways, and forces a dilemma: either accept wide and unactionable bands for what counts as normal, or use narrow bands that over-pathologize most individuals as diseased. The natural consequence of this is to start "fixing" deviations, and this is the territory of dangerous over-steering. This is why the medical community is afraid of "incidentalomas" and over-intervention. Tests which are point-in-time and compare one human to others are problematic in this respect. This is especially acute when the body is measured along many axes: like a full body scan. Much like the US Air Force Study of pilots discovered that there is no single pilot compatible with the "average" pilot measurements, comparing metrics from an individual's full body scan to a cohort average is likely to drastically overestimate problems.

In some ways, learned systems (like our physiology which is constantly adapting to our individual environment and experiences) follow the opposite of the Anna Karenina principle: every healthy individual is healthy in their own way. So the real diagnostic leverage isn't comparing an individual to an aggregate human template, but measuring change against their own homeostasis, and this requires high temporal resolution, not wider population comparisons.

I'm currently exploring how to bring this kind of longitudinal tracking to higher-fidelity modalities like wearable imaging. If any clinicians or researchers here are thinking about how to operationalize personal baselines in practice, I'd love to chat (email in profile).

Is the development experience compatible with modern AI-assisted workflows? Coding agents have drastically reduced the entry barrier for playing with software projects, and I can think of some fun ideas for Smalltalk things I've been curious about.

Why specifically DeepSeek, when there's a whole cohort of Chinese open weights models? GLM, Qwen, Kimi, Minimax, etc.

Reinforcement learning can definitely (in principle) leave this envelope, but it depends crucially on whether the feedback generates new information from reality, rather than just repackaging what humans have already discovered in the form of RL environments and densified rewards. So the question is not about RL per se, but what is the information source, what is the bandwidth of information it supplies, and then only finally the effectiveness of RL in distilling information that into models.

> Unfortunately AI excels at such problems because it oversees a much larger slice of our knowledge than most individual humans. I've always said that cross domain application of knowledge is the easiest way to make progress and I haven't really seen anything conflicting with that to date.

What is unfortunate about this? I suspect we concur in not buying into the AGI/ASI/RSI/<jargon of the day> hype, but this isn't this kind of breakthrough extremely valuable regardless?

Maybe it's not expanding the convex hull of what is possible, but the fact is that human progress is extremely jagged even within that convex hull -- for myriad reasons limitations related to funding, cross-domain expertise, number of scientists, etc.

----

If you're worried about what it takes for a novice setting out in the field, it might be useful to see how specific "mature" subfields have evolved in their training structures. Eg. I have a background in theoretical physics and it is common for grad students to do courses for a couple of more years and take a few more years to reach the same productivity compared to grad students in (say) CS/AI.

Masked emails are infinitely shareable eg. a spammer can harvest that email address and then reach you. You don't typically have a setting to block anyone else (other than the original intended counter-party) from reaching you -- unless you set up some specific filtering rule (if that's even supported).

Nice idea! JFYI, in case you ever run into design questions when extending this, you seem to be implementing a version of Object Capabilities (specifically revocable forwarders / caretaker pattern), specialized for the SMTP situation -- as you want to share with someone the capability to send you messages, but you want that capability to not be shared recursively.

Since you're also thinking about a user giving their agent access to their inbox -- any thoughts on what are the right capability primitives for that situation, and how the API & access tokens can represent that?

That's passable model assumption if your algorithm didn't create intermediate numbers/strings of arbitrarily growing length. Otherwise the thing that you're considering a constant is not actually bounded.

Whether it's an oversight in just this work, or TCS more broadly, I do not know. But it's a fundamental point nonetheless.

The string comparison model is useful in the regime where string lengths are much shorter than list lengths and can be bounded by a constant. It would be a horribly broken (i.e. unrepresentative) model if you were constructing intermediate strings whose lengths could grow arbitrarily and become much longer than list lengths.

One can of course make a puzzle out of anything, but it behooves asking what situation is being modelled by that problem.

And yes, this criticism might apply more broadly to TCS rather than just this specific artifact from OpenAI. It's reward hacking regardless of whether humans do it or LLMs.

You mean, like a really popular web service or app that someone would log into every day/week/month...?

Even easier if, additionally, many of the smaller websites were syndicated into a tracking network?

What does fiber has to do with spoilage? Is this a significant motivation for processed foods to strip out fiber?

Also, from the study:

> The non-UPF and UPF test meals were closely matched on meal weight, energy, energy density, macronutrients, available carbohydrate, glycaemic index and load, total dietary fibre, sodium and water (all ≤1.6% deviation between meals; Fig. 1b).