Yes, it's fed the score as the reward value used. If I'm not wrong, they didn't normalise it across games for the initial paper but normalised it to some range for some of the following research experiments.
The original papers [1][2] for the deep Q-network used reward clipping: "As the scale of scores varies greatly from game to game, we clipped all positive rewards at 1 and all negative rewards at -1, leaving 0 rewards unchanged".
It actually goes beyond just making you more frustrated. Venting, or catharsis, makes you feel better in the short term, creating a positive feedback loop in which you seek it out and engage in it more often, making you more aggressive and worsening the situation. (http://youarenotsosmart.com/2010/08/11/catharsis/)
Education in Asia, especially the main cities in the developed nations (Shanghai, Singapore, Seoul etc. ) is pretty much exactly the same. There's many purposes to an education, manifest and overt ones. For example, keeping youths neatly congregated and managed in singular locations, socialisation into specific social archetypes of value etc. We may say that an education is about Finding Your Passion and Being An Educated Person or to Engage Civil Society. But that's pretty much just what we're saying. Tax dollars are the votes at hand, and capitalist society really doesn't value An Education over a populace well-educated to obtain economically-productive jobs that produce tangible wealth.
We want to reverse this trend (overly competitive schooling systems) as much as we want to value philosophers and humanities scholars and artists. Enough to say it, but not quite enough to pour copious amounts of money into it. As students we're told to shoot for the sky. Study harder. Get straighter As. That's best for (capitalist) society. Every human resource neatly and fully expended. Maybe things will change when the data illustrates how people who like their jobs or areas of study are more productive. But that's a big if. The status quo seems easier.
The irony is that it's actually appallingly bad for capitalist society.
That is, it was good for capitalist society in the nineteenth century when what the economy needed was people who were literate enough to read an instruction manual but broken enough to spend all day everyday on an assembly line carrying out the same hand motion over and over again without going mad from boredom.
These days, the repetitive work has mostly been automated. The value in a modern economy is people who can think, make decisions, find creative solutions. But that requires leisure time, play, rest. So the economy is glutted with broken people desperate for jobs while it's hard to find a good manager, a good programmer or a good plumber for love or money.
I agree that schools are doing a poor job at educating for practical applications but most of that comes from rewarding for satisfying arbitrary metrics and norms that don't really correlate with real world skills. Students being motivated and competing isn't really a bad thing if they were actually learning useful things.
Historically education-performace correlation worked because it turns out that smart kids are generally good at most things so having higher barriers to entry meant higher education was a good filter for general competence and motivation.
Nowadays the "everyone gets to college" system made the filter much less reliable but still left expectations from all those kids that entered college that they will be treated as previously top percentile. Also it decreased the value of college networking as now you're less likely to meet the top percentile students of your generation.
I don't think schools were ever that good at educating for real world but they were a good signal. Leading students to believe that just by imitating the signal will lead to success is bound to leave a lot of kids disappointed.
There are a very high percentage of foreign-born parents all over the Bay Area. Last time I checked, for example, more than 30% of the residents of Alameda county were born outside the US.
One factor, in my opinion, is that for a lot of immigrant parents, their success in education enabled them to move to Silicon Valley. Education changed their lives and, many would say, in a hugely positive way.
So they apply that same perspective to their children's education and expect the kids to have the same educational experience they did.
The problem for the kids is that what it takes to succeed in the US is different from what enabled their parents to get into the US in the first place. It's not clear to me what the implications of that are, but boldness, innovation, insight, team spirit, leadership, and a lot of other desired attributes don't get discovered by students because they have no opportunity to allow them to emerge.
Even if the effect you describe is real (and I believe it is) the status quo is still easier under capitalism.
When you are poor your incentive is to find employees who are undervalued and hire them to release the potential energy in their skills. Development of those employees takes time, but you're poor. Time is all you have.
Once you have a successful business though, you have plenty of money and time is what is scarce for you. If a competitor develops your employee to extract additional value that's bad for you. If they poach your staff you have to take additional time to develop a new hire to do something the original employee was already doing.
So mature businesses shift their resources to building a moat that makes it hard for competitors to do that. This is so ingrained in how we think businesses have to work that mostly we don't even notice it. But almost all jobs contain quite a lot of "moat maintenance" tasks.
This is all, of course, depended on capitalism which says the management of resources should be done by those people who have capital. If resources were managed collectively, the moats wouldn't last because other people would bridge them. It's only under property law (or martial law) that a moat can even exist in the first place.
There's a cross platform solution built in Java, Simple Amazon Glacier Uploader (SAGU).
[simpleglacieruploader.brianmcmichael.com]
If on Linux there's a nice CLI option too.
[https://github.com/MoriTanosuke/glacieruploader]
It can actually be more X-inefficient (e.g spending indiscriminately) , because it has profits in the long run and won't go broke. I suspect the author was looking for the idea of cost, economies of scale lower average cost, hence increasing a firm's profits, not making workers more productive as he asserts.
It's very easy. Brute-forceable factorization problem. I would be interested in seeing if there were more elegant solutions though, I'm sure there's some optimizations, although the numbers multiplied are prime.
As in Plato's five regimes, democracy is only very slightly better than tyranny/anarchy. Authoritarian single party state can also be rephrased into functioning semi-aristocracy. All governments have advantages and flaws, it's better to compare them not on a theoretical demonized basis but on the output and tangible effects. For example, economically Singapore succeeds fantastically, but culturally it's missing something.
Thanks for speaking some sense. I suspect the author got the notion of Singaporean pride wrong. We're far more prideful when we feel attacked or maligned by foreigners, despite the day to day lamentations. You've stated arguments I agree with, but I do honestly feel that you're overgeneralizing to a large extent. I think it helps if we consider again what you mentioned about size. Singapore has a relatively small population. It's true, we don't produce that many groundbreaking innovations, but we need to consider the mathematical probabilities of having them in a relatively smaller population as well. My guess is that the detriment is at least an n^2 term, with less sharing of ideas, interaction and encouragement to pursue risk. As a student, I increasingly see people pursuing the entrepreneurship and more risky careers, rather than "selling out". After reading so many American articles and opinions and blogs, I don't think the idea of pursuing low risk paths is specifically amplified in Singapore. Everyone wants to do amazing things, but no one dares to. It just so happens that it's more common in other countries because of larger populations, and the environment that successful by chance entrepreneurs (I say this not to deride them but as a point of argument) have fostered. I don't have statistics to back this up, but hey we had Creative and Sembcorp/KeppelCorp. Our successes are just a lot less sexy. There's a lot of fallacious stereotyping as a result. Pragmatism doesn't imply a lack of innovation, creativity or success. On the contrary, it finds the most efficient way through to it. Our research institutions ASTAR focus on the commercialization of technology. That means we miss the sights and sounds and new discoveries along the way, we don't have Nobel Prizes, but can one really say that one type of practical innovation is better than the other?
Although props to you on the culture analogy. Nothing interesting ever happens in Singapore. For that, I think we must be at least slightly thankful. It's interesting to watch a government's politicking, gridlock and shutdown, but if and only if it's in another country
> My guess is that the detriment is at least an n^2 term, with less sharing of ideas, interaction and encouragement to pursue risk.
Absolutely. That's part of the reason for the lower level of risk-taking and innovation here. It boils down to having a small population.
> As a student, I increasingly see people pursuing the entrepreneurship and more risky careers, rather than "selling out".
Yes, this is definitely a trend, and a good one. I was talking more about the past. The pressure to go for fancy careers is still high, though - except that banking and consulting have supplanted medicine and law to some extent.
> It just so happens that it's more common in other countries because of larger populations
You seem to be arguing that it's not our fault that we have less risk-taking and innovation here, it's just our small population that's the problem. Well, I don't entirely disagree - I think our size is half the problem, and the other half is our society's historical evolution (and cultural roots). It doesn't change the fact that our market size limits our entrepreneurial prospects - unless we learn to sell to big, lucrative markets like Israeli startups did.
> we had Creative and Sembcorp/KeppelCorp
Creative was our sole shining star, and now it's essentially gone, massively outcompeted by Apple. SembCorp, PSA, Surbana and the like are ex-national corporations (IIRC) that I don't consider examples of entrepreneurship - they were built in the mould of efficiently-run government departments from the start. I do like the fact that Charles & Keith, Hyflux, BreadTalk, et al are seeing some success overseas. Still, we don't have a Sony or Samsung.
> Our research institutions ASTAR focus on the commercialization of technology
They didn't use to, and now they're driving the real scientists and big names away with their renewed focus on commercialisation. Honestly they should never have focused on basic research in the first place, but instead on commercialising technologies from around the world.
I like your optimism that things are changing - that's the spirit we need to take more adventurous paths. The more people who want to achieve big things and don't want to settle for the usual routes, the better.
To our foreign friends, our young generation is pretty alright.
Regarding Risk - what are the issues/concerns about bankruptcy in Singapore? In the United States, most of the risk is borne by the lender, and, in the event of financial collapse - the individual usually is able to come back swinging after a few years in the penalty box - which certainly reduces the downside to taking risk. If they finance with venture capital, or a LLC, then even those several years in the penalty box aren't a concern - ALL of the financial risk is borne by the VCs.
The reduction of cost-of-failure plays a very large role in whether people are prepared to risk failure.
check out the median income vs the price of housing.. the median singaporean couple can barely afford a flat with a 30 year loan.. the average couple might afford to have a baby.. or a car.. there's no room for risk for most singaporeans' lives. other than the state run lottery..check out the queues...
Two small icons next to the field, one for autogeneration (a refresh maybe?) and one for password visibility (an eye like IE uses perhaps, a concept common to other software as well.) Keep the generation code in js client side , and send the plaintext to the server via SSL as always. Normal operation of user creation. The plaintext has to be sent at least once over SSL to get hashed on the server side. Hashing on the client side is generally worse because then you'd be more susceptible to MITM attacks. (Actually, either way you'd be more susceptible. Just rely on strong SSL.)
Yes re client side generation though it might not be as random, the point is to encourage people to not use the same password. It could be combined with an educational note "Why use this password?" or "Why should I remember a different password? You don't need to..."
Client-side is always MITM friendly without certificate pinning and strong SSL, which is why I have to dial down the paranoia a bit myself: I was just now thinking that you could plant a non-random auto-completed password, but then if you can do that you could just listen for the real password too. (We're assuming you're running JavaScript here after all.)
Definitely agree. As a problem, this is pretty easy. Ask anyone that does competitive algorithmic programming. The linear solution is obvious, I wouldn't say that it requires a eureka moment, even though I didn't bother thinking of the one pass because the difference would be minimal at best (for most inputs.) OP should share what his internship was about, I'd disagree if it had more to do with understanding practical implementation and development than efficency and algorithms. Both are valid skillsets equally useful to twitter.
EDIT:apologies to whoever had to kill the extra posts, my mobile hn client is on the fritz.
