Yes. But more importantly, we are continuously training our network. There is no terminating "training set" except the set of all considered classifications. Our discovery of ourselves being "wrong" about a thing is our network continually adjusting. We also have the notion of ignorance. These classifiers are often forced to come to a conclusion, instead of having "I don't know, let me look at it from another angle" kind of self-adjustment process. "Aha, it is a cat!" moments do not happen for ai. In us, it would create a whole new layer to wrap the classifier around some uncertainty logic. We would be motivated to examine the reasons behind our initial failure, and use the conclusions to train this new layer, further developing strategies to adapt to those inputs.
HN user
theforgottenone
You really should care more about Kay has to say. There are few truly brilliant people out there, and Kay is one of them.
why use python for real work?
For reference, I was handling 30M daily records in about an hour on hardware 15 years ago. This performance is absurd.
Why does the word "product" catch people out?
reservation bailing, like scalping can be solved through new business models: for reservation fraud, start by taking a creditcard and charge a 5$ reservation fee that is deducted from your bill. For scalping, there are a lot of various approaches that work.
I only needed one.
If we need a computer science-y role model who happens to be gay, we got Alan Turing. Why get out the second team when the MVP is on the field?
Get off client coding, and shit gets more career stable.
'Sup
It is very difficult for me to translate lessons learned at google to anything except google. I have read a lot of blogs from googlers and formers who write from a perspective of, I guess, a luxury I have literally never seen, even at MS during its golden age. For example, the idea of code reviews and whitespace. I have never worked in an environment where we had that kind of money to spend.
If HP can pull this off, it will be as big a wave as the PC, the Web, and the Mobile ones, if not bigger.
And aapl and ibm.
I am guilty of this more often than I care to admit. Ooops.
or an IDE that just scans the code files, picks out the lines of comments containing "TODO" and put them in a list in the gui, so you see it all the time.
Or generate a compile warning instead of a comment. Then it will annoy you constantly, with no risk to production.
Given that we are talking about lazy/overworked on top of lazy/overworked here, why any confidence that this will be done 100% of the time?
Putting time bombs in your code, regardless of a global configuration switch, is just asking for trouble.
Diamond-square has a bug/limitation in that it creates seams. There are better solutions :).
Meh. Using a dummy object doesn't solve the problem of testing if you got the dummy or not. And throwing a specific exception is still an exception, but now more code to maintain, and only for exceptional circumstances (or is it really exceptional?).
Yeah, Null Ref Exception isn't as pretty as a more domain specific one, but the essential thing happens regardless: the program throws as soon as you try to access a member. And if you just pass your results on to another function, that function could be testing its inputs, and throwing your exception there as an explicit ArgumentNull instead of Null Ref.
I prefer less to more code, especially for an error I wont even expect to ever see. If it something I do expect, then sure, make a fancy exception and throw that.
They didn't highlight the key property of a project for when tdd works and when it doesn't: when you know the requirements with clarity (compilers, stacks, etc) it is possible to use tdd since the tests will map back to well understood requirements. But when you don't (like building a website with fussy browsers, or writing a log file parser with format anomalies), you don't have the foresight needed to write the test before the code.
I think the notion that you (eventually) get to broad code coverage in a full test pass with a single pushbutton test, that Fowler talked about, is the best idea.
I think that whether to use mocks is really dependent on your particular environment, and the penalty of not using them. I've had to mock legacy servers that I couldn't touch in testing, and they helped there a lot. Also ones with really slow IO, or bizarre security requirements that make impersonation in test difficult or impossible.
It is also a very well thought out language, that took from Java's history many of the mistakes it made, learned from them, and avoided them.
Yup, I'm bookmarking this immediately. Of course we have to go deeper, so Shit Shit NH Says is coming soon.
Eh, where was this line of reasoning when I was hand editing autoexec.bat and config.sys files to find an IRQ and DMA channel that would work with my soundblaster?
Sounds like a steaming pile of BS to me. MS could have done a far better job supporting both user levels by leaving the win7 gui alone, and adding in the start screen, and allowing this specific option during setup.
Don't believe the arguments of "individual contributor" level employees. They are lied to as much as we are.
It would be nice if they would stop promoting them too.
Hmm... funny but... why not show an example that demonstrates adding these various abstractions in useful ways. Just always printing hello world, so none of these subsequent refactors actually added any utility is funny but not as real as world. Like they could have read from a config file or command line, so the program output could change, depending on runtime conditions (but ultimately never do!) The elimination of boilerplate unnecessary comments though, yeah I love it.
Thank you for not forgetting the 101st.
What a bold prediction!
a routine that makes decisions on what action should be taken can still have a good name, like "routeOrder" or "dispatchWork". In C, inlining works if you are really worried about fn call overhead. No problem with the notion here, just... its 2014 not 1990.