Hacker News new | comments | show | ask | jobs | submit login
Statistics, Fast and Slow (timharford.com)
265 points by khc 6 months ago | hide | past | web | favorite | 40 comments



We run into similar challenges of scale and experience in humanitarian/aid work. It's easy to be overwhelmed by the numbers of suffering people when the primary dimension of representing them is something like "food consumption." Stats are vital for the logistics of meeting acute needs, but they erase (abstract out) the rich dynamics of human life.

You might think photos would help, but the endless stream of images of "starving people" ends up being cumulative, compounding the matter.

Projects that systematically dig into the lived experience of beneficiaries and treat them as complex social beings start to bridge that gap. They not only dignify the people involved (providers and beneficiaries both), but also often discover systemic issues that "the numbers" would never reveal.

I wonder what conversations the author had with people who lived there (genuine question, not a criticism).


I can't phrase this question very well, but from your experience, what are the possibilities with Virtual Reality in that? As in, could we ever reach a state of virtual-presence rich enough that it does the trick and is not the Matrix?


Good question. Or maybe in the direction of Professor Xavier's machine that (IIRC) taps him into zillions of people's experiences in parallel, but also with bar graphs? Feel everything and then zoom in to the specific?

A pretty effective approach is something like ICIJ has done https://projects.icij.org/fatalextraction/ We're working on generating those efficiently from collected data, but you still need an author of sorts. Ultimately that's still in the slide deck realm, though.

A possibly less familiar solution is what's called Executive Briefing Centers that can get pretty Minority Report-like: a wall of screens with an interactive display (mouse/keyboard/script driven). Those presentations are curated and composed (think $100k slide decks or maybe a 1-off immersive video game) to get deep into understanding how a customer segment would experience a new product in a new store (is what I've seen).

Folks like IDEO have done some experiments with 3D cameras and VR. When I've tried it, I found the non-eye-contact disturbing ("And you can't smell the shit," as a friend put it). It's one thing to not speak the language. Quite another to not know what the ritual you're observing means to the participants (the VR scene I'm thinking of dropped you into a "sacred African dance around a fire").

An important distinction here is about reporting vs interacting. We say interaction is paramount. But it doesn't really scale--numbers do. That's kindof the point of the article: both have their place. The goal isn't to traumatize _more_ people, it's to communicate--effect a sharing of worlds--with the hope of breaking cycles of trauma and violence. Like some counseling techniques, maybe numeric abstraction provides a necessary distance from the immediacy of experience that allows us to get a handle on things and make change.

Insight Share and the participatory video folks are pushing the limits with current tech. There you get into enabling folks to solve their own problems, no need to convince donors.

There's an empowering irony in using iPhones or Occuli to interview tantalum/coltan/tin miners. But I bet if we asked them, they'd have some pretty sweet ideas for where/how to direct the talent and resources being used to develop the devices to help them...

Thanks for the question!


I Was on Facebook the other day in a group about gym-going and a black lady posted a question about working out. It was evident that English wasn't her first language, so I clicked through to her profile out of interest and had a look at some of her posts.

One of them I remember was where she had laid out some new purchases or something on her bed. What struck me, though, was that the walls of her room were corrugated iron. I wish I could have sent her £50 or something right then, without sounding like a patronising twat. Not a life changing sum, but maybe it would have helped her in some way.


This TED talk [1] about seeing how the rest of the world lives was very insightful for me. Anna Rosling Rönnlund (Hans Rosling's daughter in law) went to 264 homes of varying income levels and took photos of beds, stoves, toys and other standardized goods. When sorting by income level, you can spot globalized trends by income regardless of nationality.

[1] https://www.youtube.com/watch?v=u4L130DkdOw


Oh yeah, Dollar Street (what that talk is based on) is a great project!

"In the news people in other cultures seem stranger than they are. We visited 264 families in 50 countries and collected 30,000 photos. We sorted the homes by income, from left to right."

https://www.gapminder.org/dollar-street/matrix


What about engaging her in a get-to-know-you conversation? Maybe you'd discover a different way to help (or how to send the money! :) ). Maybe she has something to offer you!


Have you read Factfulness - the book mentioned at the end of the article?

Hans Rosling has had lots of these conversations with people. They are detailed in the book as described.

It's a fantastic book BTW.


Did a quick skim through what Amazon Look Inside provides. It definitely looks worth checking out (all copies are in use at the SF pub lib).

I think I've seen Mr Rosling's presentation on birthrates before, IIRC something to the effect of: Our "first world" assumption that might drive policy decisions is that "all the poor people" are having seemingly unlimited babies, when actually e.g. Bangladesh is at or almost below replacement. Global population will peak at ~10 billion in 2050 and then decline. Africa is still a variable. (These are from foggy memory, but match what I skimmed). It's compelling and hopeful work!

The projects I have in mind are monitoring & evaluation ones that some of our customers have done. (This gets a little deeper into methods, but still extends the conversation) E.g. as follow-up on World Food Program food distribution projects: A standard food security questionnaire asks something like, "when you're out of food, do you starve, borrow, or steal?" On paper that's all you get, those 3 dimensions. A different approach had an evaluator show video clips of local people talking (scripted) about doing those things. Beneficiaries are asked to pick which video best matches what they do. Then the evaluator interviews them about why they made their choice. It opens up dialog. The best example was the man who said, "Oh, I'm fine about food. But the video I picked showed that all tress are cut down in the background. We like your food, but it would be better if you would plant trees so we could go back to farming the way we know how." ...the recommendation and ensuing follow-up was a tree planting campaign!

So, to connect it back to the article, maybe, "Do all those people want to be riding buses, living in sky scrapers, ...?"

From a different angle, that brings to mind Seth Chase's excellent doc "We Will Win Peace" (https://www.kanopy.com/product/we-will-win-peace). Looks at the "popular" perception of conflict minerals economic issues in central Africa vs reality, ensuing US policies based on perceptions, and ensuing negative impact. (BTW If you ever need a filmographer, Seth is great!)


What's interesting is when there's a big discrepancy between what the numbers say, and what the place feels like. Stats, like impressions, are not all made the same way.

When I was in Switzerland I always thought there was something sluggish about the economy, the pace that things went at, the lunch breaks, but the stats say it's one of the most prosperous places in the world.

With the Scandies I wonder how on earth the happiest country in the world could possibly be Denmark. Maybe they hide their feelings, but I found a lot of other places have people who seem happier.

Getting those discrepancies is vital, btw. You need the impressions to make you curious about the stats, and vice versa.


> other places have people who seem happier.

Happines they measure is more accurately described as "life satisfaction". There is also large differences in how different cultures express happiness. In the Nordics there is cultural tendency to downplay positive emotions and not express them publicly.

Money and Happiness Rank of Income, Not Income, Affects Life Satisfaction http://journals.sagepub.com/doi/abs/10.1177/0956797610362671

Finland Is the Happiest Country in the World, and Finns Aren't Happy about. https://blogs.scientificamerican.com/observations/finland-is...

>If happiness is the prevalence of positive emotions (let alone the displaying of them), Finland is not the happiest country. If happiness is the absence of depression, Finland is not the happiest country. But if happiness is about a quiet satisfaction with one’s life conditions, then Finland, along with other Nordic countries, might very well be the best place to live.

As a Finn, I find that I'm happiest when I scowl in the rain.


Michael Booth's "The Almost Nearly Perfect People" is a great book if the Danes' happiness is something you're interested in. A fun look at Scandinavia and their supposed perfect culture.


I'm curious, what exactly do you mean with sluggish in Switzerland (Zurich?) in comparison to which country? What I've experienced there is a conscientious, highly skilled workforce. Just small in numbers.


He compares it to the busywork, work-till-11 including weekends, BS culture most of us live in -- e.g. what many American managers cherish and measure things by.


I wish more people would realize that if you're busy all the time, you're almost by definition not working on the most important thing you could be. If you're busy it can be hard to not focus on the task at hand, which means often the big picture is lost. If you're busy, it can be really hard to explore alternatives, especially since most will be a waste of time. And when you're busy, it can be really hard to take the time to play (though this is possibly a flip side to exploring), which can be uniquely valuable.

Couple this with the engineering concept that as a system approaches 100% utilization, catastrophic failure becomes increasingly more likely, and people who are always busy may never be able to perform at their highest level. Once you fall behind, which is increasingly likely when trying to maximize short term output, it can be really hard to justify anything else. And if, god forbid, you start losing sleep, then you're going to be even more stupid, which will make any kind of catching up impossible.


>Couple this with the engineering concept that as a system approaches 100% utilization, catastrophic failure becomes increasingly more likely

Nobody cares for that, since there's nothing really critical behind most job positions -- just cogs in a bigger wheel.

In areas where this matters (e.g. people die), engineers are treated differently (e.g. at NASA).


Really excellent video on how headline statistics, such as "median household income" are often misleading and hide the real story: https://www.youtube.com/watch?v=VXI_ADnp22c

Edit: these stats are often used to push a political agenda, citing "lack of growth for the middle class," but when you dig into the stats, the story is often the opposite of that claimed by the politician, etc.


Your video argues that it's complicated to measure the growth of the middle class. But the best number shown is 23% growth over 31 years, 1979-2010. I think that qualifies as lack of growth. For reference, the GDP grew 230% in those years.

A followup video argues that lower household income is due to smaller households, but fails to mention that individual income also trends down.[1]

Though your video appears neutral, it is presented by a libertarian and the key 23% statistic comes from a conservative economist. The argument that we shouldn't trust numbers is misleading when there is well-documented evidence across a range of measures that growth is concentrating in the upper class.[2]

[1] https://en.wikipedia.org/wiki/Personal_income_in_the_United_...

[2] https://en.wikipedia.org/wiki/Income_inequality_in_the_Unite...


> But the best number shown is 23% growth over 31 years, 1979-2010. I think that qualifies as lack of growth. For reference, the GDP grew 230% in those years.

The video states this too near the end - that 23% over 31 years can also be seen as not much.

I think the main argument is that using the CPI-U or the PCE changes things significantly (from -7% to +23%). There's so many variables in any economic analysis that it's hard to trust any figure that someone shows you.


That looks like a big difference if you take the numbers out of context. But consider how much the economy has changed since the 70s: there has been a doubling of output through revolutions in technology, finance, energy, medicine, everything. Equitable growth would be 100% growth over that period.

True, there is a margin of error between different methodologies. CPI shows about half a percent more inflation per year than PCE. Similar arguments apply to measurements of climate change. They miss the forest for the trees.


Correction: the GDP grew 130%, not 230%.


I want statistics on the outcomes for people who ignore statistics, people who take statistics much less seriously than their gut, people who take gut much less seriously than their statistics, and people who ignore their gut.

My gut feeling is that the outcomes for people who use their gut over statistics are significantly better. And if you showed me statistics that they aren't, I'd trust my gut on it.


If your gut feeling was accurate, no one would be on vaccines and we'd all have smallpox, we'd only invest in feel-good companies that have no chance at profit, and communications systems (which entirely depend on statistics) just wouldn't work.

What's your point?


Actually it looks like smallpox only existed since about the time statistics was invented:

"our data clearly show that the VARV lineages eradicated during the 20th century had only been in existence for ∼200 years" https://www.ncbi.nlm.nih.gov/pmc/articles/PMC5196022/

"The German Statistik, first introduced by Gottfried Achenwall (1749), originally designated the analysis of data about the state, signifying the "science of state" (then called political arithmetic in English). It acquired the meaning of the collection and classification of data generally in the early 19th century." https://en.wikipedia.org/wiki/History_of_statistics

This correlation suggests that as soon as people stopped using their gut feelings smallpox evolved to take advantage.


The problem is that people want things to be all one way or another. The reality is that in some cases it's better to trust your intuition, and in some cases it's better to trust science. It isn't so much that science is wrong, it's just that people will try to misapply toy models often generated for a different range of initial conditions. These toy models end up producing worse answers than well trained intuition.


My gut feeling is that your comment lies within a statistically common HN grouping called taking things too seriously.


>If your gut feeling was accurate, no one would be on vaccines and we'd all have smallpox

Who said his (and most people's) gut feelings says "no vaccines"?


>My gut feeling is that the outcomes for people who use their gut over statistics are significantly better.

Statisticians preparing the stats on that will never let it be the case.


If you find this article interesting look into "The Influential Mind" by Tali Sharot. It was on FT's shortlist for Best of 2017.

The parallel is how the brain is wired, and how that effects perception / belief.

https://us.macmillan.com/books/9781627792653


I've had a chance to see stats in an ad sales pitch by a media conglomerate. Most of the stats I see are the chunks of data which are sliced to sound impressive (uplevel correlation to imply causation) Ex:- Within the region filter X, among the millennials, the sales of SUV's increased by 10% is worded as Youth buys more SUV's if you buy our ad slots with a footnote for filters

Point being, governments create misleading stats, companies create misleading stats, makes me agree even more with the author! PS:- Misleading !=(also, =!) incorrect


To be a statistician is to ignore experience. I once interviewed someone who was a statistics major. When I asked him a performance optimization question, the first thing he told me is that he'd write a program to compute the standard deviation of running time. He could've solved it in two minutes just by running the thing. No hire.


So you decided not to hire a statistician based on their intuition to first get a handle on the problem "What is the performance of the program?".

...I'd say he came out ahead in that exchange.


I think it's a pretty good idea actually, because with standard deviations (and means, of course) you can run statistical tests to make sure your improvements are statistically significant and not due to chance. I hope the candidate had a chance to explain themselves before he got no-hired.


The candidate is probably now happily working under a manager who doesn’t complain “why are you using your skills? Just [do the simplest thing that occurs to me], you idiot.”


A lot of managers out there like that unfortunately.


It also tells you how variable your performance is - for example, do you experience large fluctuations in runtime, and if so, why?


>asked him a performance optimization question

This is like asking a performance engineer to optimize the accuracy of a machine learning model, and the engineer fired up gprof. No hire.


"Producing Wrong Data Without Doing Anything Obviously Wrong!" http://www.ece.northwestern.edu/~robby/courses/322-2013-spri...


Does anybody else find that people boasting about their "hires or no hires" a) come off as showing off, b) most of times are in the wrong end of the decision?

We can comment on why we think a practice is bad -- as in, with arguments. Not about how we ditched some candidate because they followed it.


As a statistician, that has not been my experience.




Guidelines | FAQ | Support | API | Security | Lists | Bookmarklet | Legal | Apply to YC | Contact

Search: