alvis

26 ms·

301.
by alvis 9y ago
Any idea why low-background steel[1] is that valuable given the fact that it is only 0.005 mSv/yr above natural levels at 1.26? [2] [1] https://en.wikipedia.org/wiki/Low-background_steel [2] https://en.
302.
by alvis 9y ago
I disagree with the point that "there's no alternative to crafting a well-structured presentation". I'd argue that a jupyter notebook or equivalent (e.g. pweave) is 1000X better than powerpoint.
303.
by alvis 9y ago
No doubt, the content is the essence of all presentations. All tools are just aids. As you suggest, we have been perfecting presentation for thousands of years. To an extent, powerpoint has captured some of the best ways of communication in
304.
by alvis 9y ago
"As yet, no new vision of personal computing like the one that came from Xerox PARC in the 1970s has emerged." Despite many attempts, namely keynote, prezi and slide.js, the way we communicate remains largely the same. Can't
305.
by alvis 9y ago
As an Imperial's PhD and having spent couple of years doing research there, I can say that big salary is not essentially the main factor. Certainly it is a factor, but probably not as big as Pantic thought. The most important factor is
306.
by alvis 9y ago
In fact, it is perhaps the best way to avoid you joining a place you will dislike, as they won't let you in to begin with.
307.
by alvis 9y ago
Babel can't do type checking, right?
308.
by alvis 9y ago
It really depends on the application. Certainly, everyone love control, but there is a trade off between control and cost. Firebase prediction is simply a simple implementation for some simple use cases, which are plenty.
309.
by alvis 9y ago
TypeScript is never mean to be a replacement of Babel. The spirit of TypeScript is code integrity. Imagine that you are working on a project which change often. In this case, refactoring and maintaining breaking changes are headache. How wo
310.
by alvis 9y ago
Another option is pnpm( https://pnpm.js.org ), a lesser known but much faster alternative.
311.
by alvis 9y ago
Sadly they failed the HN DDoS test.
312.
by alvis 9y ago
Shall we talk about Deepo, not docker? Absolutely loading loads of packages in a container is never a good idea for production. But Deepo is not for production, it is a playground. It is a thing which help people to explore and learn these
313.
by alvis 9y ago
Out of curiosity, I am wondering if entrepreneurs really worry too much on their competitors. Often I find articles saying the opposite -- they urge people not having a blind eye on their competition so that they can at least differentiate
314.
by alvis 9y ago
"Do things that don’t scale" -- The essential of the essential IMHO. Intuitively, it sounds nonsense. After all, who what to build an unscalable business. But having been working on my super early stage startup for a while, I real
315.
by alvis 9y ago
I cannot agree more on that having a simpler and clearer pattern is perhaps what we need. Solving 90% of GraphQL use cases with 1% of the complexity is a killing point to me.
316.
by alvis 9y ago
This is likely untrue. Unless you fetch data only from one single database, so that you can perform an atomic transition, GraphQL cannot guarantee you to receive the data you want. Technically, a GraphQL server may combine data into a singl
317.
by alvis 9y ago
As a GraphQL user, hardly I can say it is perfect, but it does solve many current challenges arisen from the surge of mobile use, namely the waste of bandwidth on headers and unused content, as well as latency between requests. From the ser
318.
by alvis 9y ago
I am keen to second here. With a PhD in probability & loads of experience in data analytics, my experience has told me that we are too ignorant and sometime too ambitious to try to predict a the outcome of a stochastic process (e.g. fin
319.
by alvis 9y ago
I, as a very organised perfectionist, am so keen to have one. However, having said that, while it would be a great tool, I'm just not quite sure how many are willing to pay for it. In my opinion, a tool providing insights (e.g. basic d
320.
by alvis 9y ago
In fact, it is a new problem in fin tech. While ML can mimic human decision quite accurately nowadays, machines remain largely incapable to handle new information. Now suppose again we have a machine that rate load applications. In the past
321.
by alvis 9y ago
Can't say our machines are trained based on facts, but purely data. What you get out of it is from what you feed into it. The challenge we are going to face in this AI era is that the training data can be biased or include some morally