Want to read Slashdot from your mobile device? Point it at m.slashdot.org and keep reading!


Forgot your password?

Comment Re:Never (Score 2) 369

Yup. Same, though I'm more on the software side of things.

AI is great for tasks with well-defined input and output, especially if they are tolerant of significant errors. But a huge fraction of engineering simply isn't like that. For the most part, AI will just become another tool in software engineer's toolboxes that will be brought to bear when it makes sense, and other solutions will be used when it doesn't.

Comment Intelligent animals are unlikely (Score 3, Insightful) 158

There are many animals that are quite intelligent, but it's exceedingly unlikely that we would ever have the capability to genetically engineer an animal like a pig to have a human-like brain, even if we wanted to.

The reason is simple: our brains are way bigger than pig brains (human brain: about 3.5lbs, pig brain: 0.4lbs). In order to have a pig with a human-like brain, you'd have to completely reshape it's skull, and because a pig skull is very different from an ape skull, you'd have to do it in a very different way than humans do. To do this, you'd need to generate a wide variety of novel adaptations to make it so that a pig can support a brain that's about 8 times the size. That's just not happening.

What is being done in these kinds of experiments is far less ambitious: to use small amounts of human DNA to make animal tissues compatible for transplantation. This kind of research has gone on for a long time: it's common to genetically engineer mice and rats to have human immune systems, to make them better test subjects. In this case, if the research continues, you'll have a pig growing a pig liver, with some of its genetic markers changed just enough to fool a human body into thinking that it's a human liver rather than a pig liver.

Comment I don't buy it (Score 1) 292

The problem here is that the federal government just doesn't have much control over many of these regulation, as they are mostly down to state and local regulations. At the federal level, then, implementing these policies as Pai describes isn't just about removing regulations: it's about increased regulations of the states, something that Republicans have been historically against, and which may open the way for court challenges.

I also worry that he explicitly mentions private industry, but doesn't appear to care about the deployment of public broadband (an effort which has begun in many communities, but is often stymied by state laws implemented as a result of telecom lobbying).

My big worry here is that these claims are a smokescreen. He doesn't actually want to regulate states more strongly. He wants to remove federal regulations related to the Internet, such as the 2015 net neutrality rules. For some of the things that Pai has opposed, see here. If he follows previous patterns, his effort will decrease competition, increase prices, and not make the slightest dent in broadband speeds.

Comment Re:NOT Fake News (Score 1) 372

The memo may well have been intended to include peer-reviewed publications, however. My interpretation of the USDA's later statements is that this is the interpretation of the law that is actually legal: government employees have fairly broad free-speech rights, and by law cannot be silenced unless it's regarding classified information.

I fully expect Trump to violate this law left and right, and I'm genuinely worried that he will largely succeed.

Comment Re: Continuation of the Bush policy "Hear No Evil" (Score 1) 372

That's a rosy view that isn't justified. True, Trump is worse than Bush, and Bush was worse than Reagan. But Reagan's administration was also pretty invested in hiding the truth, especially when it came to economic policy. His administration is the one that started the whole "trickle down" economics canard that the Republican party has been pushing to this day, and that policy was always pushed using lies about it creating growth.

The downward slide of the Republican party started back with Nixon and his contemporaries in the party, whose southern strategy resulted in the Great Southern Realignment: the Republicans, starting with Nixon, shifted to a political strategy of exploiting racial prejudices to win votes (before this both parties were roughly equally-racist). This shift in political strategy, away from good governance using conservative principles and towards drumming up fear and tribalism, created an environment ripe for a party whose platform does little but enforce the desires of the wealthy, as it does today.

Reagan may not have been as bad, but he was a major factor in the Republican party we know today.

Comment I doubt this is good for China (Score 1) 120

A big part of the cause of the 2007/2008 crash was too much financial innovation (mortgage-backed derivatives, specifically). This may make help China's economy seem to be growing well in the short-term, and will almost certainly make some people rich, but it's very likely to blow up in their faces.

Comment Re: A new fad? (Score 1) 339

Sorta kinda. It's not just a fancy name, for one, but an entirely new way of defining types.

Interfaces in C++ are also far more limited than concepts. For one, interfaces only work on class types. Concepts work on both classes and built-in types (e.g. int, float, pointer types). To make use of interfaces, you must pass objects by pointer or reference, not by value. Classes used this way have to be virtual, which carries a performance cost. There's no clear way in C++ to state that an object adheres to both interface A and interface B: you'd have to create a third abstract interface class which inherits from A and B, and have your class inherit from this third class. Interfaces are also highly non-standardized, which would mean that in order for interfaces to replicate the functionality of concepts, library developers would need to come to an agreement on a common interface set and then completely refactor their class definitions to conform to that common interface set.

So yes, it is possible to refactor code to use interfaces and virtual classes everywhere to produce a similar design to the one offered by concepts, but it will require a lot of work, won't be terribly readable, and comes with a performance penalty.

Comment Re:Considering is different from doing something (Score 1) 218

The big evil from premature optimization comes from attempts to optimize the code that don't impact the computational complexity.

My usual path to writing non-trivial code is:

1. Find the lowest-complexity method of implementing the problem that is still readable.

2. Write the lowest-complexity method in the most readable way I can, even if it adds operations that may lower performance but don't increase the complexity (e.g. extra copies of variables, redoing simple computations).

3. Re-evaluate the above if performance proves to be an issue down the road.

Of course, sometimes the optimal method is far more complicated and less readable than the simpler method with higher complexity. In these cases I try to balance performance expectations with readability and code writing time in a reasonable way. But I've found that the majority of times I've had to go back and fix poor-performing code has been when I (or the original author) has slacked off on computational complexity, so I try to avoid making this shortcut whenever possible.

Comment Re:Scary **** (Score 2) 292

In the abstract, I think it could be done in a way that would be helpful but with minimal danger, but it wouldn't be easy.

The key would be to construct the system so that information coming from another vehicle is never trusted. The information might still be useful, but only in terms of refining estimates gathered from the car's own sensors.

Comment Re:Good for them! (Score 1) 858

I think this rosy view of corporate competence paints too pretty a picture. People very often do get to the top more based upon who they know and how many resources they have, rather than their accomplishments. Peoples' accomplishments in corporate environments matter to some extent, but usually don't result in any dramatic role changes.

The main check against incompetent people reaching upper-level jobs is that businesses that allow this to happen often don't last very long. Tellingly, Trump has had an impressive string of failures in his past, and his personal wealth has likely grown less than it would have if he'd just invested it in an index fund. This also demonstrates that when you have a lot of money it's very possible to get around any competency checks. Furthermore, if a company doesn't have much in the way of competition (which is increasingly the case for many US corporations), then it takes egregiously bad behavior to bring a company down, removing competence checks on upper management further.

That said, I think you're spot-on with regards to government. I'm sure it doesn't always work, and internal cultures of various agencies can sometimes have problems that persist unchecked. But I think that overall, these represent fairly modest inefficiencies in the system. If Trump destroys this system, it will cause lasting damage to the US government for decades to come. I'm sure Republicans are salivating at the thought.

Comment Re:Good for them! (Score 1) 858

There are some significant protections that prevent many of these people from getting fired at a whim. For now, at least. Trump would have to significantly change the rules by which these organizations act internally, and that would probably involve a significant number of lawsuits.

In the mean time, there's zero value in having people in the organization who are sympathetic but unwilling to act to counter Trump's policies. There is a word for such people: collaborators.

Comment Ridiculous (Score 1) 255

There's no way that Trump's policies will substantially increase inflation, for a number of reasons.

The biggest reason is that it's very easy for the Fed to reign in inflation by increasing interest rates, and by all accounts a Trump presidency is likely to try to push the Fed to be more aggressive about doing this than the current Federal Reserve Board.

The next biggest reason is that Trump won't really be doing much of anything to increase spending. Trump and the Republicans will very likely blow the federal deficit wide open, but this will primarily be through tax cuts to the rich, which have very little impact on inflation.

I'm pretty sure that we can expect low inflation and decent economic growth over the next couple of years. But Trump, by removing financial regulation, will make another crash like the 2008 crash much more likely. When that next crash happens, then Bitcoin and gold will very likely spike in price (it'll be impossible to predict precisely when: sooner with Trump than it would have been with Clinton, but it could still be 5-10 years away).

Slashdot Top Deals

What hath Bob wrought?