Logical uncertainty Posted on Feb 20, 2015


I recently have come across a quote by Alan Turing which I found very illuminating: The view that machines cannot give rise to surprises is due, I beliieve, to a fallacy to which philosophers and mathematicians are particularly subject. This is the assumption that as soon as a fact is presented to a mind all consequences of that fact spring into the mind simultaneously with it. It is a very useful assumption under many circumstances, but one too easily forgets that it is false.

Read More

New platform Posted on Feb 20, 2015


I am setting up a new blog using hugo to generate it statically. I have already imported some of the old posts I made from in my old posterous blog, “Rought Insights”, and I will be importing my posts from Distributed Republic as well. I will try to blog almost every day. My goals are currently to keep a thought journal, and to improve my writing in English. The blog is tentatively named “Probabilistic Pontifications”, because alliterations make for good blog names.

Read More

Vague Questions Posted on May 11, 2012


I hate it when people ask you a question and then stop in the middle, leaving the meaning of the question – and more importantly – the assumptions behind the question, open to interpretation. I hate it because it’s lazy and it’s rude. It offloads the task of precisely formulating the question by filling in the missing pieces to the person you’re asking. I tend to get very frustrated. Most of the time I will give a short yes or no answer.

Read More

Anticoagulants when needed Posted on Apr 21, 2012


A tiny sensor could monitor blood pressure and delivers anticoagulants if it spikes up. It would help avoid nasty strokes without the downsides of taking anticoagulants daily. Questions: If a clot blocks a vessel, how fast does the pressure wave travel through the blood system? It would depend on the flexibility of the blood vessels I suppose. If the anticoagulant is released then, is there enough circulation to make it reach the clot?

Read More

Blind experiment Posted on Feb 29, 2012


Double blind studies are the gold standard in medicine. I contend they should be triple blind. The statistician analyzing the data should be given anonymized, whitened, normalized data from the experiment and know nothing else about it. The experimenters can then see if the model takes their variable of interest into account. This particularly applies to epidemiological studies.

Read More

Phrasebook Posted on Aug 17, 2011


Hello, I regretfully do not speak your language, but I memorized this sentence to impress my friends. Would you kindly play along and act as if I just told you a funny joke? We could even have a small conversation, I will speak gibberish and you can act as if it makes sense. Thank you very much!

Read More

The computation estimation tradeoff Posted on Jul 22, 2011


Probability distributions classically represent ignorance (at least for orthodox bayesians). I’ve longed wondered if and how it could be applied to represent thinks which are knowable but hard to compute. Here’s a rough insight to formalize this! Instead of calculating the result of a call to a function, one can return a probability estimate. The error is then lowered by progressively doing more computations. This is a form of lazy evaluation since computation is deferred, and it allows one to get useful results while avoiding long computations.

Read More

Reputation networks and branding Posted on May 12, 2011


I can think of a couple things that branding achieves Establishing a history of good relationship with customers. Being a costly signal of trustworthiness. Tap into tribal instincts (hello Apple). The two first functions are a low tech way to build reputation. Reputation networks could efficiently replace branding and strongly diminish the economies of scale. There are still economies of scale (the more customers, the stronger the confidence in your brand) but the scale power coefficient is likely much smaller.

Read More

Alphabet Posted on May 3, 2011


hinese writing objectively sucks. The alphabet is a superior invention, and so is writing horizontally (field of view is wider). How obvious is an alphabet though? Apparently it’s not! All existing alphabet (except some designed fairly recently and possibly meroitic) seem to lineate to the proto-sinaitic alphabet. There is no sign of independent discovery. Scripts as distant as Khmer and Runic share a common ancester. It probably means that over periods of centuries, languages had no alphabet, until they came in contact with an offspring of the proto sinaic alphabet.

Read More

Vacation office Posted on Apr 27, 2011


A nice perk from a company would be to offer its employees to work part time, one month a year in a vacation office. An office / beach resort in the carribeans for example. A strong case must be made for perks, as they are generally inferior to a higher salary. There are strong reason to provide this perk though. It’s not a benefit, the company just happens to have a paradisiac office.

Read More

Connect. Socialize.