2019 Week 9: Optimisation and Rowing

Short version: Life is full of metrics, and applying optimisation is useful, but be mindful of why not merely the how. Rowing is big in Oxford.

Long version:

Optimisation:

I can be a little obsessive about optimisation. I think it comes from doing a lot of mathematics. In the upcoming book The Metric Society, Steffen Mau describes a world where increasingly scores can be assigned to all aspects of life. I would guess the most common metrics people are concerned with are their bank balance and their weight, usually trying to maximise one and minimise the other. In everyday life I often find myself crunching numbers at the supermarket, aspiring to shave seconds off my running pace, cramming one extra experimental condition into the day. Is optimisation ultimately worth aspiring to? I’ve been reminded by friends this week that life isn’t always an optimisation problem. Optimisation can help me get faster, or spend more time at an art gallery, but it can’t tell me which one would make me happier, or is a more worthy way to spend time. I think it is important that I remember the ever growing pool of metrics are tools, rather than ends in themselves. More on this another week.

Rowing: Torpids 2019

This week the Isis in Oxford has been busy with the shouts and splashing of students rowing races. Oxford is relatively famous for its rowing. The Boat Race between Oxford and Cambridge along the Thames attracted between 250,000 and 400,000 spectators in 2017. However unlike that race, or in Olympic rowing, the boats in this week’s races (known as Torpids) do not compete in parallel, but rather line up in sequence and chase one another.

Introduction to bumps racing:
Oxford is made up of many colleges which compete in, amongst other things, rowing. Since the main river through Oxford is narrow, and the race is contested by many boats (73 men’s boats and 61 women’s boats this week), it is not possible to race side by side. Instead they compete in divisions of 12 boats (13 boats per race including the top boat of the next lower division), with each boat starting simultaneously but separated by intervals of 130 feet. The aim of the first boat is to finish the course without being “bumped” by a boat further down the sequence, whilst all other boats are aiming to catch the next boat ahead. In Torpids specifically, once you “bump” (by overtaking, making physical contact, or the boat in front conceding) you leave the race, whilst the boat that has been bumped carries on. This protects the boat that achieved a bump from then being bumped in that race, even if the boat immediately behind would have eventually (or imminently) caught it. How this plays out is that in each race a boat usually only maintains its rank (known as rowing over), moves up one position (bumping), or moves down one position (is bumped), and so only the top four seeded boats meaningfully contest the position of top ranked boat (known as “head of the river”). Getting a bump on each of the four days is nearly always the best performance a crew can hope for, and is rewarded by “blades” (oars decorated with the names of the crew) being earned. Since a boat that has been bumped carries on rowing, it is possible to be bumped multiple times, and this is reflected in some pretty significant drops (e.g. Pembroke 2nd boat falling 8 places on the last day of racing). Further complicating the results calculations is that a boat that is bumped can continue on to bump, which is the only circumstance where a boat can gain multiple places. This creates the incredibly unlikely possibility that a boat can rise from the bottom of it’s division (the 13th boat) to the top (1st) in one race: only achieved by each boat bumping the one in front in the specific sequence of reverse order; 13 bumping 12, 12 bumping 11, and so on until having been bumped by 3, 2 then bumps 1. This would result in the boats then being ordered in reverse order to their starting order. Assuming then performances consistent with that first race, all boats would then proceed to row over (finish the course without being bumped). Notably this system of racing strongly encourages starting as hard as possible to catch the boat in front as early as possible.

Notable Performances of Torpids 2019:
Teddy Hall Men’s 1st boat were the highest seeded crew to get blades, gaining four places in Division 1, whilst GTC Women’s 1st boat also got blades in division 1, with an even more impressive 5 places gain. Brasenose Men’s 1st boat gained the most places on their path to blades at 7, while Hertford Men’s 2nd boat had the biggest fall of an impressive 14 places.

More details:
You can see the full results of Topids 2019 here. The full Rules of Racing are here. St Hugh’s Boat Club provide a Glossary of Rowing Terms here.

Upcoming Rowing Events:
Oxford and Cambridge compete again in London on Sunday 7th April 2019. Bumps are next being raced in Oxford on 29th May to 1st June 2019.

Other Rowing Stuff:
Australian Rowers Eat 25-30 MJ per day, British Rowing, Rowing Australia, RowingNZ,US Rowing.

Photos from the Week

2019 Week 8: 2018 Altmetric Top 100

Short version: I write about the top 21 highest impact papers of 2018 according to Altmetric, a rank based on mentions in academic publications, news media, and social media. They mostly relate to health or the environment.

Long version:

Metrics and Rankings

It is interesting and useful, but difficult, to be able to compare the quality and impact of publications. There are many methods to quantify this, one of which is Altmetrics. This scores publications based on mentions in the news, on Facebook, in patents, and other sources which are tallied and weighted. Since it takes less time to write a tweet than a citing publication, Altmetrics respond much more quickly and reflect a much wider potential audience than more conventional measures, but for the same reasons are less accurate measures of quality.

My brief thoughts on the top 21 papers ranked by Altmetrics from 2018:

1. Mortality in Puerto Rico after Hurricane Maria

The US centric nature of the metrics comes out clearly with the top ranked paper being about deaths in Puerto Rico after Hurricane Maria. President Donald Trump was not good to Puerto Rico. The study measured 4645 excess deaths from 20th September 2017 to 31st December 2017, 70 times higher than the official toll of 64. Deeply disturbing.

2. The spread of true and false news online

Fake News travels faster online than truth . Words like clickbait, chainmail, and memes matter a lot more when presidential elections and lynch mobs become an issue. I am reminded of this early Tom Scott video about flash mobs.

3. Alcohol use and burden for 195 countries and territories, 1990–2016: a systematic analysis for the Global Burden of Disease Study 2016

I am disappointed to have missed this paper in Week 5. This article from The Lancet recommends the best level of alcohol to consume is none, finding alcohol causes deaths through tuberculosis, road injuries, self harm, and cancer. There are some clever methods used to look at measuring actual consumption by individuals, presented in maps on the third and fourth pages.

4. Trajectories of the Earth System in the Anthropocene

This is pretty grim. Describes the “Hothouse Earth” as “…a pathway that could not be reversed, steered, or substantially slowed”. Talk to people about climate change. Work out what you can do to minimise your impact. Katharine Hayhoe has a great facebook page that can help bridge the gap for climate change skeptics.

5. Association between physical exercise and mental health in 1·2 million individuals in the USA between 2011 and 2015: a cross-sectional study

Exercise usually makes people feel better, and the biggest rewards occur going from none to some. While your brain does use a significant proportion of your daily energy expenditure, the rest of your body needs movement to function properly. This study quantifies that relationship with mental health, finding optimal benefits from sessions of about 45 minutes and 12 to 20 sessions per month.

6. Dietary carbohydrate intake and mortality: a prospective cohort study and meta-analysis

Studying nutrition is made difficult because the timescale of the effects you are trying to measure (over lifetimes) means that experiments are difficult to control. That said this study finds that ideally 50% of your energy intake should come from carbohydrates, and if you cut carbs, you should not supplement them with animal fats and protein, but go for plant based options.

7. Evidence that the Great Pacific Garbage Patch is rapidly accumulating plastic

Just because information is painful or disappointing, doesn’t mean it should be ignored. This paper described the exponential growth of plastic in the Great Pacific Garbage Patch. This SMBC comic is highly relevant.

8. Complementary Medicine, Refusal of Conventional Cancer Therapy, and Survival Among Patients With Curable Cancers

The death of Steve Jobs is a famous case of “Complementary and Alternative Medicine” demonstrating that brilliance in one area is not mutually exclusive with stupidity in another. This paper points out that cancer patients who opt into these unsubstantiated treatments are more likely to refuse more established medicine, and therefore are twice as likely to die. Relevant Tim Minchin beat poem.

9. Global warming transforms coral reef assemblages

The Great Barrier Reef is undergoing a series of bleaching events. While bleaching does not guarantee death, the longer the bleaching the harder it is for corals to recover. This study quantifies that relationship, and finds a nonlinear relationship where prolonged or intense heat results in rapidly increasing losses. More recently a million tonnes of sludge is to be dumped on the reef. The SMBC comic is still highly relevant.

10. The biomass distribution on Earth

The biomass of humans is approximately 10x that of all wild mammals, and half that of livestock. Plants rule the world.

11. Radar evidence of subglacial liquid water on Mars

Reminds me that NASA has a sense of humor too. Useful to know if you believe in Terraforming Mars.

12. Comparative efficacy and acceptability of 21 antidepressant drugs for the acute treatment of adults with major depressive disorder: a systematic review and network meta-analysis

Depression can be treated medicinally, but I have an (uninformed) suspicion the diversity in response to treatments reflects a diversity in the underlying illness. This study compares antidepressants.

13. The impact of the ‘open’ workspace on human collaboration

Discussed in The Economist, this paper from Harvard Business School finds open plan offices seem to counterintuitively decrease face to face interactions and increase online communication. Having worked in a few different office layouts, I would suggest that the staff themselves make a bigger difference to communication than the layout. Also sociometric badges are a thing.

14. Structure and Distribution of an Unrecognized Interstitium in Human Tissues

Exciting to see a microscopy paper in the top 20. Highlights the importance of checking assumptions and going back to the fundamental structure, rather than abstracting.

15. Risk thresholds for alcohol consumption: combined analysis of individual-participant data for 599 912 current drinkers in 83 prospective studies

Although being slightly more positive on drinking than the alcohol study above, particularly noting “increased alcohol consumption was log-linearly associated with a lower risk of myocardial infarction”, overall the study found a decreased life expectancy overall for alcohol consumers. It goes on to suggest lowering the recommended limits for alcohol consumption to less than 100 g per week.

16. Death or Debt? National Estimates of Financial Toxicity in Persons with Newly-Diagnosed Cancer

The US continues to confuse the rest of the world with its views on health care. Specifically this study reveals that of US cancer patients, 42.4 % will have used up their entire life savings within 2 years.

17. Effect of Low-Fat vs Low-Carbohydrate Diet on 12-Month Weight Loss in Overweight Adults and the Association With Genotype Pattern or Insulin Secretion

It doesn’t seem to matter if you cut down on calories via lowering fat or lowering carbohydrate consumption, so long as you’re consuming less you will lose weight. Particularly interesting is the undermining of genotype evidence.

18. Weaponized Health Communication: Twitter Bots and Russian Trolls Amplify the Vaccine Debate

It’s interesting to question the causation here: are Russian Trolls trying to lower vaccine acceptance to weaken health in the US, or are they trying to build rapport with people prone to unfounded conspiracy theories, such as being anti-vaccine, to then promote other (political) falsehoods. Also: how vaccines work.

19. Decreases in global beer supply due to extreme drought and heat

While climate change is bad for both natural and agricultural plants, if the previous two alcohol papers are considered perhaps less beer in the world is not a bad thing.

20. Primary care-led weight management for remission of type 2 diabetes (DiRECT): an open-label, cluster-randomised trial

I’ve had several friends in medicine mention their direct observations of this in individual patients, but here it is presented as the aggregated results of 306 individuals. Weight loss results in remission of type 2 diabetes.

21. Association of Coffee Drinking With Mortality by Genetic Variation in Caffeine Metabolism

Coffee (maybe) is good for you! But the study suffers from “a ‘healthy volunteer’ selection bias”. Interestingly “Participants drinking 4 or more cups per day … were more likely to drink instant coffee and be current smokers, whereas participants drinking 1 to 3 cups per day were older, more likely to have a university degree, and more likely to report “excellent” health.” So the causation question remains: does coffee make you healthy, or do healthy well educated people drink coffee?

Extra Thoughts

It’s a little unsatisfying to be able to spend so little time on such interesting questions, but the nature of inquiry is a trade off between breadth and depth. This week I opted for breadth, and it serves as a reminder for how much interesting content there is being generated in the world.

Photos from the Week

It’s warming up in Oxford, and Toxic Daffodils are blooming. I managed to fit in watching a Cuppers game at the historic Sir Roger Bannister athletics track. Port Meadow is flooded, and the reflection in the sun is stunning.

Writing from home.

2019 Week 7: Zero to One

Short version: I read Peter Thiel’s notes on startups, had a couple college formal dinners, cycled in the Cotswolds, and bought some protein powder.

Long version:

Zero to One

Peter Thiel is founder of PayPal, and an early investor in Facebook, which have made him a billionaire. In 2012 he taught a course about startups at Stanford, which via Blake Master’s notes became the book Zero to One.

I will update this post later with my thoughts on the book and other missing sections.

Photos from the Week

2019 Week 5: Alcohol

Short version: I’m drinking a bit less, the media still likes skewing science to get page views, live music is wonderful, it snowed more in Oxford.

Long version:

Alcohol

Personal Experience:
Last week I wrote about having made the change to not eat meat. More recently I’ve been toying with cutting back on alcohol. Throughout January I observed a one drink per day maximum. This was most difficult socially, as I don’t have a single reason for cutting back, and there are many contexts where accepting drinks is the polite behaviour. Moreover it often felt more difficult to explain setting a threshold of one, than simply saying I didn’t drink, or was undertaking Dry January. I think in general this comes from a desire for principled consistency. While consuming alcohol is fun, I felt overall I was able to have just as good a time while consuming less, with benefits felt in sleep and recovery from training.

Guidelines:
The Australian Government Department of Health recommends no more than two standard drinks per day long term and no more than four standard drinks per day (where a standard drink is 10 g or 13 mL of ethanol). The British NHS has a similar 14 units per week, to be spread over 3 or more days , but using smaller units (8 g or 10 mL of ethanol). Of course the US is a little trickier to compare due to the lack of the metric system but the CDC recommends one drink for women and two for men, with much larger units (14 g being 0.6 ounces, or 18 mL).

Benefits of Alcohol:
In February of 2018 Dr Claudia Kawas presented at the American Association for the Advancement of Science Annual Meeting in Austin, Texas her work “The 90+ Study“. The quote from her talk “I have no explanation for it, but I do firmly believe that modest drinking improves longevity” led to headlines such as “Drinking Alcohol Better Than Exercise For Living Past 90“, “Study: Drinking Alcohol More Important Than Exercise to Living Past 90“, and “Drinking Small Amounts of Alcohol May Help You Live to over 90, Claims Study“. New Zealand joined the party almost a year later with “Drinking wine better than exercise if you want to live a long life“. YouTuber Doctor Mike did a comparison. After searching for a couple hours though I couldn’t actually find a peer-reviewed paper on these benefits. The study itself places less emphasis on the alcohol vs exercise question; “People who drank moderate amounts of alcohol or coffee lived longer than those who abstained.” (which is kind of funny since consuming alcohol and coffee seems to be pretty dangerous.) Notably the only mention of ‘alcohol’ or ‘exercise’ in the AAAS official news post was “Data show that the risk of developing dementia has declined slightly in the past decades, Kawas said, which she attributes to people improving their lifestyles: eating better, exercising more, trying to minimize stress.” I suspect omitting the alcohol point is deliberate.

Other Alcohol Stuff:
WHO Global status report on alcohol and health 2018. Dry January does not lead to increased drinking in February. People are drinking less in pubs (on-trade) and slightly more from supermarkets (off-trade).

Music: Soloists and Symphonies

I listened to the City of Birmingham Symphony Orchestra perform Prokofiev’s The Love for Three Oranges: Suite, Beethoven’s Violin Concerto, and Vaughan Williams’ Symphony No. 4. I particularly enjoyed the drama of the final piece, and the incredible energy of the orchestra. The soloist Augustin Hadelich has a pretty incredible life story, and incredibly clean sound.

Stuff I read this week

The Australian Bureau of Statistics National Health Survey 2017-18 suggest some fairly grim things about the diets of Australians. AI is getting really good at video games.

Photos: 5 Things that made this week great

Writing from home.

2019 Week 4: Vegetarian

Short version: Why I don’t eat meat, some things I’m reading at the moment, and three photos from my week.

Long version:

Eating Less Meat

Since having moved to the United Kingdom, I’ve been eating less meat. The strongest motivation comes from a desire to minimise carbon impact and waste. The change has been made easy by meat substitutes being widely available and cheap (and delicious).

Evidence:
Intuitively, in order for humans to gain nutrition from an animal, that animal needs to consume either other animals or plants first. Given these processes cannot be completely efficient, there is always a lower energy cost to the environment to eat lower on the food chain. The real effect of this is described in the chart below.


High Protein Diets:
The CSIRO published a book advocating a high protein diet. It generated some controversy in part due to being funded by the meat industry. It still seems a high protein diet is healthy, but there is evidence swapping animal protein for plant protein lowers mortality overall.

More reading:
Publications from Nature and World Resources Institute about the impact of meat on climate change, which include the infographic above. Also, from a moral point of view, Consider the Lobster.

Stuff I read this week

Relevantly, restaurant reviews from the blog Vegan Eats Oxford. The Graduate Outcomes Survey was released. Matt Levine continues to write humorously about finance. Hybrid Perovskite Semiconductors are cool.

Photos: Cakes, Climbing, and Snow

Writing from home.

2019 Week 3: From Notes

Short version: I had a busy week, but made note of a few interesting things I came across.

Long version:

From Conversations

Empty trains run to Berlin’s empty airport, summaries from Bloomberg in 2015, The Economist in 2017, and BBC in 2018.

It is possible to live relatively normally without a cerebellum, a condition known as cerebellar agenesis.

There are techniques to treat mice to make them transparent, Nature News discusses this paper.

Life is more enjoyable if you make an effort to appreicate things. Example: rather than suffer through doing your laundry, enjoy that you are fortunate enough to have items you like.

From my Inbox

Google:
In 2018 I visited 150 places in 45 cities in 4 countries travelling about 35,000 km. I posted 106 photos to Google Maps which were seen 182,107 times.

Headspace:
Advice on New Year’s resolutions

Strava:

FinTech

Riding the Tube, I noticed the surge in banking start ups, particularly Viola Black running an advertising campaign referencing Monzo. It seems their SEO game is a little weak, as a google search for them has the top result being from Monzo forums. It makes sense that with Monzo and Revoult being valued over $1bn, you would see a surge in competitors, but the market seems a little saturated?

Jack Straw’s Lane, Oxford

I broke my bike last week, and had to run to pick it up this weekend. Discovered some new routes around Oxford.

2018 Week 27: Mileage and Musings

Fitness:
At the end of this September I plan on running the Blenheim and Oxford half marathons, ideally setting a personal best. Last year I ran the Blenheim Palace race in 1:47:13, and ambitiously I would like to do better than 1:30:00. I suspect this is too large an improvement to expect even with 3 months to train, but it provides a target to work towards an incentive to train harder.

Oxford has some beautiful places to run. Below are some pictures from routes around Christ Church Meadow, Port Meadow, and Oxford Parkway.

Small Holidays:
This weekend I visited Oxford’s Arboretum, and Blenheim Palace. It was lovely to enjoy the summer sunshine. With so much interesting content so easily accessible online and on paper, it can be difficult to prioritise taking time to just experience the outside world. Emotionally, it felt very valuable to be out somewhere a little different, to take a small holiday. Some photos from the weekend:

 

Car Ownership:
It has now been two out of a relatively few visits to the palace that have included a car show, and this time it was the Pre ’50 American Auto Club Rally of the Giants. It is remarkable the strength of the relationship people form with their cars, and this is particularly noticeable in the context of a community that defines themselves by the restoration and care of classic machines. The reasons that come to mind for this strength are: cars are one of the largest purchases commonly made, cars provide freedom (of movement) to a unique extent, and for many commuters cars are a often occupied second home. I enjoy the practicalities and aesthetics of cars (and motorcycles, and bicycles), but for something so common, the economics don’t seem to add up. Intuitively, for such an expensive piece of infrastructure, the majority of (privately owned) vehicles seem to spend a rather minimal amount of time being used. While I (like many) enjoy driving, it would seem the more hours I spend driving my car, the fewer hours I would have to focus on other tasks, and as such relocating or finding alternative transport (even if slower) would net gain me more hours of useful time. I suppose it would be interesting to quantify this from the perspective of homo economicus, but a quick search of the literature give me only “semi-structured interviews with 19 regular private car commuters” and “discrete choice models of the household’s decision to own zero, one, two or three or more vehicles“, i.e. interesting descriptions of individual and collective behaviour with regard to owing cars, but not a judgement on if it makes sense.Quotes

Some interesting comments include “Drivers frequently fail to appreciate the full costs of their travel and equate running costs with fuel costs only” and “Thus the autonomy and empowerment drivers feel can benefit health and wellbeing and access to a car is associated with superior physical health, less depression and lower mortality rates”.

I suppose, ultimately, it would require pricing relatively difficult concepts such as the flexibility of point to point transport to an individual, or the sense of safety it provides, or the value in transporting dependants.

Writing from the office.

July 2017: Job Hunting and Rock Climbing

I left Sydney for Oxford in early July, so a month having elapsed, this is what I’ve been up to:

Job Hunting
My primary goal has been finding work. So far that looks like:

   Positions Considered             81
   Applications             16
   Rejections               6
   Recruiter Calls               4
   Interviews               3
   Offers               0

Unfortunately no offers. I’ve focused on jobs in Oxford, so have applied to positions from the university. In a wider search, reed has seemed more relevant to me than indeed,  and making a profile there has been the main source of calls from recruiters, which seems promising. It’s probably time to extend the search to include London, though I’m still hoping to find something here in Oxford.

Reading
Aside from many job descriptions, I’ve been reading Huffington’s The Sleep RevolutionThe Economist, and reddit.

Hackerspace
Having become a member of the Oxford hackerspace, I’ve been helping out a little with the 3D printing service. Producing tangible objects is very satisfying in contrast to job hunting.

Fitness
I’ve tried two new activities: squash and rock-climbing/bouldering. My staples of running/cycling and weightlifting leave a gap over hand-eye coordination, which squash fills in well. After about 8 hours I think I’m starting to develop muscle memory for basic play. I was surprised to find climbing has a huge mental component; planning and executing the correct strategy seems just as important as strength. It’s certainly fun, and much more mentally involved than lifting weights or running, but for pure fitness training not as effective.

Chess
I’ve played 207 games of chess. I haven’t improved much, so should add structured training rather than just blundering repetitively through similar openings. It’s probably also been more procrastination than relaxation this month.

Berglabs
And, of course, I’ve resuscitated this domain and updated it with a WordPress theme.

Travel
Life has taken me to London (by train) and Eyam (by car).

Writing from the Clarendon Laboratory, Oxford.