adyus

It appears there's a Javascript SDK listed as available.

Is it better to have reduced fraud and disgruntled customers, or viceversa?

I'm genuinely curious. I suppose it depends on the ultimate goal, but seeing as Coinbase is VC-funded, well...

I don't remember where I first read it, but I liked this opinion: in the future, decentralization, even if only adopted by a techie minority, will serve as a threat of what could be, thus keeping the centralized powers in check.

That's true, but I cannot recall a recent instance of "cooked books" since everything started moving to centralized databases. The blockchain keeps looking like a solution in search of a problem that could be solved easier by other means.

> It is recorded at some date.

This is where the human element comes into play, and something the blockchain cannot secure.

It's still buggy, but it mostly works as intended. I've been using it together with https://github.com/badaix/snapcast/ for whole-house multi-room audio.

Nice! Here's an interesting idea:

You could use a grayscale height map and apply textures based on relative or absolute height. Tallest point on the map? Must be a mountain. Cluster of points lower than the mean or median? Could be a lake or river.

Essentially, it would emulate real world feature distributions.

Note to OP: you just got several hundred dollars in quality advice from a very experienced source. Talk about lucky!

Would transfer learning help with that?

So, https://home-assistant.io then?

You mean Mattermark?

It's all manual. Take a look at what Dashlane is doing (essentially writing a plugin for each site).

Here's what I could find in a couple minutes:

https://github.com/openai/universe-starter-agent

OpenAI's example universe agent. Remember that while their goal is an agent that works in any and all environments (read: games), you could certainly optimize yours just for MarioKart.

Congrats on finishing the project! As you've already linked at the bottom of your post, it's possible that OpenAI could've solved most of your I/O issues.

One thing I'd suggest is exploring a reward function, instead of using only pre-recorded training data. That is, give the AI a goal to complete (in this case, finish the race) and let it learn by itself!

Design is empathy. That is, trying to answer the question "How can I make this easier to use or understand for someone else?"

So the answer is yes, we could all use more empathy.

Hey Stavros, I'm curious about your lighting and motion detection setup. How did you get the lights to fade in and out? Is it purely in software (meaning the switches are permanently on)?

Are you using PIR for motion detection? Did you layer any logic on top of that to keep lights from turning off if you're stationary for too long?

I'm currently working on that as we type: http://ezsend.it/

This is strictly for one-time shipments, targeted at the occasional sender. There are better solutions out there for small businesses.

"Material Design" and Google's Material Design are different things.

The former is a label generally applied to flat, square design with minimal flourishes. The latter is a set of principles and standards for how digital interfaces should behave, along with a sample implementation.

In Google's MD, the clickable items are supposed to be identified either by their elevation (drop shadows) or an all-caps, secondary-color treatment and position at the bottom of elements.

It's true, icons and some text links could use some improvement.

Well, that content has to take some form. Previously it was all about skeuromorphic design, to make things look like their real-life equivalent.

Then, the balance swung to the other extreme, with completely-flat design. Things were decidedly 2D and "digital".

Material Design is Google's attempt (among others [1]) to standardize while providing subtle cues of depth and interaction.

[1] - https://www.lightningdesignsystem.com/

It looks like the first (hero) element on the page breaks Material conventions. You have a photo with a large z-index somehow below a panel with a smaller z-index.

The idea of Material design was not just pretty shadows and drip animations on click. It was about providing a metaphor for digital content on screens.

Still, great job on the technical effort of converting everything to Material!

The beauty of (NLP) bots is supposedly that you can just ask what they can do.

Whether that's a viable UX strategy is still up for debate.

Looks like a nice tool for a DBA's arsenal. It would be more valuable if it also showed the impact of each change, where possible, in terms of performance, security, etc.

In other words, it would add more value if it could translate the actual events into human concepts of impact.

Well, if one has to input a lat/long pair to get "3645 OUR DUDE - Central,Paraguay", the same effort[1] can lead to better data in Google Maps or better, OpenStreetMaps

[1] http://gis.stackexchange.com/questions/29079/my-house-addres...

I think it's been mentioned in the original XAddress post, but the problem is not necessarily mapping a lat/long pair to a human-memorable address.

The way current addresses have worked (mostly) is by associating with landmarks. Large street names are known by lots of locals. Smaller streets are associated with larger streets by proximity (second left on Large Street). Numbers help place an address at the beginning, middle or end of a street, on the left or right.

Grid cities have an added advantage of encoding approximate location in the street name directly.

XAddress and What3Words solve a problem for the 4bn without street addresses, but they do not help locate things or navigate to them without being online.

I don't think that's the point of the project, rather decentralization and breaking free from proprietary data silos.

The point is that you decide where your data sits and who has access to it, not that it's free-for-all.

Unless I misunderstood your point completely...

I think the purpose is making the latter as easy to remember and recite as the former.

A small side note: I checked the footer for an "About" or "Team" page, but found nothing. Checked the "Contact" page and sure enough, it's an East-European company.

There seems to be a fear of publicly disclosing who is behind projects in the whole area. Ondrej, could you offer an opinion?

How does that mitigate against the one machine being stolen?

The convention is that the chances of two devices getting stolen together are smaller. That's why offsite backups are a thing.

Generally, payment processors give you a token once a client gives them a credit card. You just need to abstract away the processor-specific code that uses that token.

Or the complete opposite: some people can read bash/chef code faster than a long, wordy post.