Two Lines on Everything
[Part 1]
the cardinal [Richelieu]... was in the habit of saying that with two lines of a man's handwriting, an accusation could be made against the most innocent
I enjoy thinking about things, but am consistently frustrated with how difficult it is to let others know my thoughts. I feel like I need to record my thoughts somewhere, and I might as well do it here.
This is not a claim to authority; it is an admission and acknowledgement that the words here are open to criticism. I do not claim to be the most innocent man, nor the most honest man, nor the best thinker. I have contradicted myself in the past. I have lied, bent the truth to my own purposes, and redirected conversation. Take these words with a grain of salt. Feel free to respond to me, either on Twitter, or via other means, and I will attempt to present criticisms together with my thoughts.
This is the second part of my thought dump. You can call it a blog, if you like. I sometimes call it that. It has been split into multiple parts because loading times are getting longer. Other parts can be found here:
Two Lines on Everything, Part 1
2024-Feb-25
Regarding Crab Claw Alternators (RMS #2223), presumably there could be a disc-rotor version of this, with a ring magnet in the middle, and metal bent in such a way to collect the magnetic flux and distribute it above and below a ring of coils. I suppose an issue is the winding of the coil rings, which would be a fairly tedious affair.
2024-Jan-26
Game idea: Pore Miner - you are a nanometre-sized diver, tasked with swimming into a PromethION flow cell to work out what's gone wrong with the flow cell. The flow cell is actively sequencing, DNA and other contaminants are blocking the pores, and the carrier ion concentration is reducing as the run progresses. Can you save the flow cell and maximise run yield, or will you be zapped into next week as a failure?
2023-Dec-25
You may have heard of Simpson's paradox:
Simpson's paradox is a phenomenon in probability and statistics in which a trend appears in several groups of data but disappears or reverses when the groups are combined.
I just discovered that it applies to vaccination statistics and COVID mortality statistics in Aotearoa. I was inspired to write the following in response to this reddit discussion:
See the Vaccination Details table at Te Whatu Ora. Looking at categories as a proportion of the total, I get the following comparative statistics (not broken down by age due to privacy reasons):
-
No doses: 3.55% of recorded total; 9.95% of hospitalised total; 17.36% of ICU total. Demonstrating that people with no vaccine doses are overrepresented in hospitalised people, and even more overrepresented for people in ICU. These statistics are, I think, the best way to "slice the data to prove vaccination is a significant factor."
-
Fully vaccinated (7 days prior): 27.04% of recorded total; 17.54% of hospitalised total; 17.26% of ICU total. Demonstrating that people who are fully vaccinated are underrepresented in hospitalised people, and similarly underrepresented for people in ICU. These statistics support the reasoning I have about vaccine effectiveness, i.e. vaccines stop severe disease (as designed), but if someone gets too ill, the vaccines won't help their recovery.
-
Booster (7 days prior): 57.48% of recorded total; 59.55% of hospitalised total; 56.02% of ICU total. Demonstrating that people who are boosted are overrepresented in hospitalised people, and underrepresented for people in ICU. This oddity (i.e. apparent effectiveness of the vaccine flipping for hospitalisation vs ICU) is possibly due to the confounding effect of age.
There is an age and vaccination status for deaths table further down. Statistics around age and deaths are more complicated, because there are two possible comparative totals that might make sense to use, i.e. the proportion of people in the general population who have died, and the proportion of people who have died within 28 days of getting COVID. You can fetch the 2022 general population proportions and mortality statistics from StatsNZ:
- 0 to 59: 75.66% of population, 54.77% of total deaths, 8.43% of deaths after COVID
- 60 to 69: 11.20% of population, 4.76% of total deaths, 8.86% of deaths after COVID
- 70 to 79: 7.79% of population, 8.32% of total deaths, 20.61% of deaths after COVID
- 80 to 89: 3.56% of population, 14.06% of total deaths, 36.17% of deaths after COVID
- 90+: 1.79% of population, 18.1% of total deaths, 25.93% of deaths after COVID.
A few observations: the group of people under 60 years comprise the majority of deaths from all causes (unsurprising considering its a much larger group of people), but a comparatively small proportion of deaths within 28 days of getting COVID. Older people who have died are more likely to have died after getting COVID than not.
Unfortunately, the vaccination status age breaks in the Te Whatu Ora tables and figures don't match their mortality and hospitalisation statistics, which makes comparisons challenging. What is apparent is that there is a trend of increased vaccination among older people. Eyeballing the graph, here are my guesses for proportions of unvaccinated people:
- 0 to 59: 10-15%
- 60 to 69: 7%
- 70 to 79: 6%
- 80 to 89: 5%
- 90+: 4%
So older people are overrepresented in vaccination statistics, and also overrepresented in deaths within 28 days of getting COVID.
It gets interesting when comparing these proportions to the COVID mortality statistics. If vaccination status had no effect, then the proportions of dead people who are unvaccinated shouldn't change from the above proportions in the demographics breakdown. However, here are the "not fully vaccinated" proportions among people who have died within 28 days of getting COVID:
- 0 to 59: 24.83%
- 60 to 69: 14.1%
- 70 to 79: 11.46%
- 80 to 89: 9.35%
- 90+: 9.63%
In all cases, these proportions of unvaccinated people are higher than the proportion of unvaccinated people in the general population. In other words, unvaccinated people are overrepresented in the proportions of people who have died within 28 days of getting COVID (about double in all cases).
This is classic Simpson's paradox. Because the 70-90+ age groups are such a huge proportion of deaths within 28 days of getting COVID, the statistics of "not fully vaccinated" proportions are skewed towards those groups, which ends up being less than the "not fully vaccinated" proportion in the general population.
2022-Aug-26
I just saw this tweet, and it reminded me that this "Maths is about the process as much as the solution" grated on me.
The reason it grates on me - the flash of insight I had today - is that almost every other teaching experience doesn't care about the process. Maths is the only area where markers care about how I solve a problem, rather than that it is solved at all. My frustration is that the process, my process, is argued about. In life outside school, people also (for the most part) care about end products rather than the process.
I have a hypothesis why Maths is this way: people are presented with canned problems that have perfect solutions, which means that the answers themselves are boring.
... and here I am writing this, and listening to someone talking about psychology and human behaviour... which is also about processes rather than products.
Understanding and improving processes is important and useful. People have different ways of thinking, and it's not wrong to think in a particular way that fits our physiology and life learning. In a commercial sense, the process of production should matter. We should care that water was extracted from an underground reservoir underneath a parched country, and packed inside crushed and compressed sea sludge. There are often multiple paths to a solution, some short, some long, and some that lead to the wrong solution. We should care more about how things are created: the tools that were used, the inspiration that was used, the whakapapa.
Okay, maybe you do have a point, after all.
Regardless, I will attempt to cling to a shred of dignity around my frustration in the design of maths problems. Where abstract discrete maths is important (e.g. counting small numbers of things), then it makes sense to have discrete problems. Where it's not, it shouldn't be squeezed into a perfectly-fitting box.
2022-Jul-13
I understood the Alert Level system. It made sense to me:
- All cases controlled, no community risk
- All cases are controlled, but some risk of community spread
- Some community spread, but should be manageable
- Uncontrolled community spread
https://en.wikipedia.org/wiki/COVID-19_alert_levels_in_New_Zealand
If we were still in the Alert Level system, according to those guidelines all of Aotearoa would be at Alert Level 4. People would be staying home unless they were doing essential things (e.g. grocery shopping, keeping society from being destroyed).
But the government wheels started to come off once we got our first untracked community Delta case in August 2021. New Zealand shifted to a split alert system in September 2021, and stayed that way for a month.
https://en.wikipedia.org/wiki/Timeline_of_the_COVID-19_pandemic_in_New_Zealand_(2021)
"No, we can't keep the country at Level 4. The infections are only in Auckland. We'll shift to Level 3, with additional restrictions."
On October 4th 2021, the government decided to ignore expert advice, and tried to put fences around infected areas:
"No, we can't move the country back to Level 3. The spread is only in Hamilton. We'll keep the country in Level 2, with additional restrictions in Hamilton." #PP
I now have my own rationalisation on why the government may have changed their approach on October 4th 2021: they stopped believing the experts, who were saying something along the lines of, "Trust us, if you hold for a few more weeks, it will get better." The experts had been saying that for over a month, and while it was getting better, infections weren't disappearing completely. We'd have to suck it up sometime, and Auckland was getting antsy. After all, it wasn't that bad. Only a few cases.
It didn't work. But by that time, the wheels had been set in motion. It took years to build up that trust, and only a month of strained relationships - from a country divided - to break that trust:
"No, we can't keep Auckland and Hamilton at Level 4. It's been too long. We'll shift them to Alert Level 3 step 2, i.e. Alert level 3, with additional restrictions."
Excuses are easy to find to justify almost anything. It doesn't really matter if the experts don't believe them. What's important is saving face; doing what makes economic sense in the short term, even if it causes more problems over the long term. All you need to do is convince the people who already support what you're doing. The new Traffic Light system was beckoning: a papering over of Level 2-like restrictions to make it appear like things were changing while staying the same. Besides, they didn't want to disrupt people's Christmas plans for that silly new Omicron variant:
"No, we can't go back to the alert level system. We've been planning this shift for months, and the Omicron case is only in flight crew. We'll keep the traffic light systems, and hold in red until just after Christmas."
https://en.wikipedia.org/wiki/COVID-19_Protection_Framework
They had a vaccination target. Increasing infections were trumped by better protection for non-vulnerable populations:
"No, we can't stay in red. We want to give people a reward for vaccinating at a high level. We'll go to orange, because we think we can manage this."
There was a brief burst of insight from within the government; the only time it seems to me that countrywide restrictions have been increased since October 2021:
"Omicron is in Nelson! We're going to red, everyone!"
But that was short-lived. The government decided infections and deaths had dropped to an acceptable level:
"No, we can't stay in red. It's been too long; we'll shift to orange, with additional restrictions."
And so it continues, even as cases, hospitalisations and deaths rise above that previously-acceptable threshold:
"No, we can't go back to red. We're pretty sure we've got this. We'll stay in orange, with additional restrictions."
Meanwhile, I live on a knife-edge, sharing concern with many other people from Aotearoa, and staying home as much as I can. I need to manage my own risk, within a country where many people are much more willing than me to take risks, and will only change if the government tells them to do so.
The country should be at Alert Level 4. It should have stayed that way since August 2021; a couple more weeks back then, and we would have nailed this. I would be doing more productive work, worrying less about what my future holds, and whether my kids would still be living in fear of ongoing, more destructive infection.
Today I've heard that a new Omicron variant, BA.2.75, has a new name of "Centaurus", provided by a random Twit, and boosted by Ed Yong. It seems as long as we stay within the Omicron branch, there won't be any more official Greek names. It continues to amaze me how much powerful people dislike community-driven change. I shouldn't be amazed by that, but I am.
2022-Jul-09
Living with COVID.
* shudder *
It sucks that our government expects everyone to fend for themselves, with their COVID-19 advice being essentially, "brace yourself for a harsh winter and further infections."
In the hope that I could spark some debate and discussion, I'd like to discuss what I'm doing.
First, some disclaimers. I'm in a rare and privileged position that I can work almost entirely from home, going in to work at night every week or so to archive emails, check on reagents & consumables, and do 3D printing.
I also enjoy physical solitude.
Your mileage may vary.
We are living within a worldwide pandemic for a respiratory virus. My primary goal is to reduce the amount of air I'm breathing from people I don't normally live with.
If there's no shared air, and the primary mode of transmission is through the air, infection can't happen.
I don't go out of my house to interact with others unless it's necessary, or I think the risk of infection is low enough that the interaction can be done safely.
I'm comfortable if I'm masked up and meeting a few people outside. Anything more than that, and I get antsy.
When I leave the house, I wear a mask. Whatever I feel comfortable with. I consider the mask to be the last line of defence.
At the moment, I have been mostly re-using some N95s that I was given while on Jury service a couple of months ago. I've got a few that I cycle through.
When I meet people, I check my surroundings to see if I can improve ventilation. Can we move to somewhere with high ceilings? Are there windows I can open a tiny bit? Are there doors that can be opened? Am I separated from other people?
My partner and I go to the farmers' market once a week, usually alternating whose turn it is. It's an outdoor market, with lots of people (many others wearing masks). I consider it a small infection risk; it's probably the place I get exposed to the most COVID viral particles.
I am vaccinated (Pfizer, currently 2x doses + 1x booster). I don't consider the vaccination to be full protection; it makes me immune, not invulnerable. My body takes time to amount an immune response. Vaccination helps me to build up a defence quicker, and recover quicker.
I have an undiagnosed chronic respiratory issue, which gives me low-level cold & flu symptoms all the time. This makes it difficult for me to know when I should be testing for COVID-19. I have decided that if I experience symptoms worse than what I consider normal, then I test.
We have a supply of RAT tests that were provided free by the government. My chronic COVID-19 symptoms (most frequently headache, runny nose) mean I can easily pass the low bar that the government has for tests: one less barrier I need to worry about.
I have a partner and kids, all going to the same small school. We are more cautious than the other parents with kids at the same school. We test when the kids get a cough or a sniffle, and keep them home until they've recovered.
We are privileged enough that we can do that.
To help us through our colds, we keep our house warm, with windows slightly ajar, and use cough lozenges.
There are a few different lozenge brands that have active ingredients that have demonstrated antiviral effectiveness. Codral is one of them:
https://www.mdpi.com/2673-8112/1/2/37/htm?_param=763284718
When I first heard about COVID-19, I hunted through Google Scholar for randomised control trials for coronavirus antivirals. Something that kept coming up was carrageenan. So I bought some Irish Moss cough syrup and some nasal spray:
https://journals.physiology.org/doi/full/10.1152/ajplung.00552.2020
We have a room in our house that we can use to sort-of quarantine one person, and we've used it once after a suspected exposure event.
All RAT tests so far have been negative (we always do throat swabbing as well as nose), but I expect we'll get it eventually.
In the event of a confirmed infection, we'll start wearing masks around the home, especially for the known infected people. Just a few extra barriers to reduce the amount of viral particles floating through the air, and give our bodies a fighting chance of getting through this.
I don't know for sure if any of this is the right thing to do, but I have confidence that the sum of it is at least better than doing nothing. And the more we can do to reduce the spread of infection, the better it will be for all of the society we live in.
[this is a collection of my Twitter thread on this matter]
2022-Jun-27
Two things that go really well together: PCR tubes and passive swivelling.
I remember a video, many years ago, of someone carrying out very fast PCR using a stepper motor that flipped an arm between a bath of hot water, another bath of less hot water, and a fluorescent primer to help work out when amplification was complete. I was amazed at how quickly the PCR was done, with just a small change of the reagents (and possibly polymerase). These days, I'm still amazed, at the speed of it, but also at the lack of interest that researchers have in carrying out high-speed PCR.
More recently, I found out about a 30 minute PCR SARS-CoV-2 amplification method using a machine called NextGenPCR. I wondered if the company had considered using KAPA2G or SpeedSTAR, as those polymerases can achieve super-fast PCR cycling time. One of the company representatives replied to my question with this Tweet, about a PCR machine that did a similar thing involving a PCR plate moving forward and back along a skate, and heat blocks moving up and down via a piston.
Just tonight, I was thinking again about the process, and wondering if these two concepts could be combined in a better way. The problem I saw with the NextGenPCR machine is that it is very big, and the moving parts have both a lot of mass and a lot of distance to move. Moving large chunky things is a problem with 3D printers as well, and is the reason why people sometimes prefer bowden tubes to direct-drive extruders.
In any case, I realised that reducing both of these problems could be done by moving the tubes as little as possible, and shifting the heat blocks so that they end up heating the tubes in essentially exactly the same place. An eight-strip of PCR tubes could rest inside a swivel bucket with a lid over the top of it, and the heat blocks could move along linear rails to radially engage with the tube, moving along rails that intersected with the swivel point. I'm imagining quite a small little contraption, possibly with bevelled or chamfered holes at the top of the heating blocks to encourage the tubes to swivel into the right position without any additional active rotation. The tubes could be flanked by two sensor "tubes", which would measure the internal temperature to get a more accurate idea of how long the tubes had been at the desired temperature.
I'm currently thinking that a lock/release mechanism for the heat blocks might work well, somewhat similar to those rocket pens with multiple colours, where each block being pushed up releases the previous block, and it sits passively in the engaged position without any additional energy. Such a mechanism would allow a single rotation-to-linear punch to be used for engaging each block, simplifying the mechanics and therefore reducing the number of things that could go wrong. Also, the blocks could be heated via induction, meaning that no wires would need to be connected to the blocks at all.
I intend to have a go at animating this with either FreeCAD or OpenSCAD, but wanted to write this out first before I do any further research or experimentation, just to see how close I can get to a final solution via the first thought experiment. Also, the depressing nature of experimentation can sometimes lead to "this is never going to work, so I shouldn't try it" approaches; this will at least remind me of what I was thinking at the time I first thought of the idea.
2018-Oct-03
[This entry has been retained / duplicated here, because I think it's important to understand what prompted me to begin this thought dump]
In September 2018 there was a Senate Judiciary Committee hearing, in which we were introduced to an excellent candidate for the Supreme Court of the United States. A person who spoke with humility, who understood the power of truthful statements, and the need to be aware of different perspectives. This person was well-read, demonstrating an incredible understanding of human nature, and had an established record of comment on a number of important sociological issues. While they didn't have a commanding knowledge of the entire history of law, their career alone is evidence of a great ability to read, learn, and understand.
This person was not "one of us"; they appeared to be elevated above the usual behaviour of society, and it was hard to imagine they would not have compassion for the suffering of someone who appealed to the Supreme Court for justice. They demonstrated their own sacrifice: a selfless act of shame and humility in order to improve the outcomes of society as a whole.
Unfortunately, this person, Professor Christine Blasey Ford, was not the appointed nominee for the Supreme Court of the United States.