HN user

madisfun

90 karma
Posts0
Comments45
View on HN
No posts found.

Consider what happens to a soul which always gets what it wants.

The soul craves for what's not easily attainable.

Commercial transactions are all the same.

Feasibility is sweet but one-dimensional.

There're things you can buy, and things you cannot.

You cannot buy a different self.

Politics, volunteering, social games, arts and sports

are the new frontier.

And we act to define what we are.

The companies like 23andMe can do anonymous tests. They can even sell "gift cards" for cash (or bitcoin, for digitally inclined). Don't deliver kits, let them be picked up without asking for an ID, just by the gift card number. As long as the test results are not linked to a specific address or name, it's fine.

They will still have clues about the state and the city and the IP address of the person. It's enough to do whatever research they may want to do with aggregate data. It's not good enough to sell this data to insurance companies and advertisers.

GDPR requires consent for the processing of the personal data. Displaying an ad per se is not regulated by GDPR, and does not require consent. Though personalization of this advertisement requires processing of the PII, and thus it is supposed to be regulated.

Facebook UI is full of dark patterns, but it is possible to withdraw consent through settings: www.facebook.com/ads/preferences/

What it shows is that it's risky to rely on a "cloud"/subscription service offered by a company from a different jurisdiction. In particular, if the service provider is a US company, it's a red flag. If it's Chinese, it's a red flag too.

Round-robin and privacy do not dwell well together. Like mike-cardwell pointed out in another comment, it just distributes the same information to more parties.

As there has to be at least party which will know the request, some information will be leaked. But what can be prevented, is giving "unrelated" requests in the hands of the same resolver. Few of the request per se are interesting, the combinations of them allow to build user profiles.

The policy should not be round robin, but somehow based on the domain itself, so that all requests about the same domain go to the same resolver, but to nobody else.

An even better mechanism would take into account who is the owner and the controller of the domain. So that requests about, let say, facebook.com and fbsbx.com land at the same resolver, but github.com and microsoft.com by another.

I'm not sure if transferrable and sellable rewards make CCGs less of a gambling activity. If anything, obtaining a physical object which can be exchanged for cash, with a potential for profit, is the definition of gambling.

Non-transferrable rewards exploit the same weakness of the human psycology, but offer even less in return.

Both kinds of gambling in games should be regulated. It's not easy to define what constitutes a lootbox, and what kind of randomized output is safe. The industry will probably argue that the lootboxes are actually randomized content, not prizes.. From that point of view they sell a non-transferrable experience (service), not a tangible item (product).

Living in Europe, I have yet to see a smart navigation app, which would allow to take into account:

- tolls and fuel consumption (Via Michelin is great at offering 2-3 routes with different and realistic costs, but the app is an ad-nest and their navigator is bad; Here can avoid tolls but the UI is bad; Google Maps is useless if you want to optimize the costs).

- altitude changes (a marginally shorter road which has to climb and descent 1000 m / 3000 ft is not the same as doing extra distance on the flat ground).

- the shape of the road, prefer wider and more direct roads (there are many narrow and extremely curvy roads that can't be driven fast, and some may be very stressful to drive).

- the realistic median speed as an estimator (not the speed limit; sometimes it is set unrealistically high; also do not assume that all tarmac roads are equivalent).

- local and temporary traffic restrictions (respect car-free hours in some areas and restrictions on a particular kind of vehicle or engine).

- eventual stops during long trips

Google Maps is far from an ideal navigation app at this point. The only reason it is used, IMO, is the abysmall UI experience in built-in car navigators. (Anecdotal evidence: I rented a car last week, it took me five minutes to enter destination, and then I couldn't figure out how to interrupt navigation. Eventually just used the phone to drive around).

Tl;dr: Dear Google, Waze, etc., distance is not everything.

Might be a good idea to have a throwaway SIM and a phone for travelling, not logged in into any cloud or social network services, and not tied to your main google account. Maybe just a couple of chat apps to stay in touch with your friends/relatives. Switch SIM back to your main number and reset phone once you're beyond the border... It seems exactly what EFF recommends by the way. With a laptop it's more complicated.

For example Intel's C++ compiler produces (or used to produce?) much more efficient numerical code than either gcc or clang.

I'm not an expert, but I believe that Intel could have implemented their hardware-specific optimizations in any other compiler framework (either gcc or clang). In this case multiple language implementations, while commercially viable, are not beneficial to all users.

Mersenne Twister has a huge state, and if you have to use the MT (you need a very long period), you will probably want to be able to initialize it properly. For the common use case is a linear congruential generator which is initialized with just one integer.

Google's Roboto is a very close drop-in replacement for Helvetica. Almost the same metrics, slightly more "condensed"/DIN feel.

You may also consider FreeSans (GNU replacement for Helvetica) and Liberation Sans (which is a metric replacement for Arial, which is an alternative for Helvetica...)

Also, what's wrong with Open Sans?

On paper, the results are amazing, but researchers are always biased to produce positive results.

BTW, a match between StockFish 10 and LeelaChessZero, an open source implementation of the same idea, will be organized in a couple of days. From the LC0 blog:

Lichess.org will host a match between the mighty Stockfish 10 and Leela. It will be a 6 games match with time control of 5'+2" with ChessNetwork commentary. Games will be played on 15th December at 17:00 UTC.

Stockfish 10 will run on 64 cores 2.3GHz Xeon, while Leela will use the latest v19.1 Lc0 with 11248 network and will run on one GTX 1080 Ti + one RTX 2080 GPU.

I believe it was TCEC season 13 which put GPU engines in disadvantage due to overheating, it was also the first season with GPUs. Right now there is TCEC season 14 in progress, did I miss anything?

It would be nice if A0 participated in at least one public computer chess championship, Chess.com's CCC or TCEC. That's a level playing field and all games published.

AlphaZero was a great concept and execution, but if we have to judge its relative strength, it should compete fairly. 4 TPUs (~ 4 Titan V) + 44 cores for AlphaZero vs only 44 cores for Stockfish pre-9 may or may not have put Stockfish at a disadvantage.

BTW, current, presumably balanced, TCEC 14 configurations are:

Non-GPU Server: CPUs: 2 x Intel Xeon E5 2699 v4 @ 2.8 GHz, Cores: 44 physical, RAM: 64 GB DDR4 ECC

GPU Server: GPUs: 1 x 2080 ti + 1 x 2080, CPU: Quad Core i5 2600k, RAM: 16GB DDR3-2133

TCEC GPU server looks more modest than what A0 authors used to "beat" SF.

If I did the math right, then if the notebook I typing this comment on mined at 100% CPU load for 1 hour, the payout to the script owner would be equivalent to US$0.00085. For nearly 200 Wh consumed, or 70g CO2 equivalent, or $0.004/kWh. That's two orders of magnitude less than the cost of the energy consumed, supposedly somewhere around 3% efficiency.

I suppose asking for donations/selling subscriptions/products/whatever is more sustainable. Even if 3 users out of 100 agreed to donate/pay the monetary equivalent of 1 kWh (~ $0.140), the site owner would receive more than forcing 100 users to heat the air for one hour and putting 7kg CO2 in the atmosphere.

The web doesn't need browser-based mining. The web needs affordable and convenient microtransactions. With something better than a 0.05 + 5% fee.

In almost all wars, the "bad" is the group that lost the war. And when it cannot be decided or doesn't matter anymore, those who control the discourse decide who is/was "good".

Taking something away is not good or bad per se. Sometimes both sides see the other part "taking something away" from them (see any rebellion vs central power conflict).

Finding ocean is easy: water and gravity have formed the surface to have few local minima. It would be much more interesting finding peaks: there are many local maxima.

And the most important maxima are often surrounded by lots of local maxima. They may also have some steep slopes so gradient methods can easily overshoot. Sometimes the surface function is not even continuous (overhanging walls).