nxb
Regardless, US culture says that this sort of thing gives the employer far too much power over the employee, and that there are plenty of unscrupulous employers who use that leverage against the employees to make them do what they want, far beyond what the US considers appropriate.
EEU culture typically says the world is far worse without these books and that there couldn't possibly be any downsides that outweigh the costs, or that there are bandaids which could theoretically fix any problems with it.
Neither side will likely ever convince the other side on this topic, so no point in arguing about it :)
These slavery books are still pretty common in parts of Europe too.
You're in for a MASSIVE culture shock, if you ever get employed there. For this, and many other reasons. The cultural expectations in business are often the far opposite of that of the US.
Don't forget that individuals and group culture are two distinct things, but both are very real things. Let's be clear about which one of those we're talking about.
Sure it's probably all obvious things, but it would be nice to have a checklist to skim over, to be sure I haven't forgotten any major consideration when designing a new system.
If I'm reading correctly, it seems they tried to deduce the bottom chart from the top chart, which is not appropriate for the reason you wrote.
The top chart only considers ONE year into the future from TODAY, based on someone's age TODAY. More specifically -- It's a chart for survival analysis (predict rate of failure, based on a certain fixed time period and a certain starting date) not life expectancy (predict the time period over which a certain fixed rate of failure occurs, based on any future date.)
The bottom chart though, should be far more influenced by the predicted increase in life expectancy -- since it would have to account for decades of predicted average life expectancy improvement, instead of just increase in average predicted life expectancy over one year, which probably has a negligible effect on the top chart.
The reality is more fluid and less black and white.
> Second, after our consultation and looking over the specifics of your situation with my team
Sounds like you're calling out the specifics of his project as being problematic. This vague yet specific line is going to attract questions and / or arguments.
> it looks like we're probably not a good fit for your project."
= The "it's not you, it's me" method. A very obvious and overused way of saying that you don't like something, but you'll pretend that there's something wrong with you. Even young children can see through this attempt at controlling emotions.
Regarding the length, it's extremely common for people to interpret long excuses for avoiding something as deceitful, even if that's not truly the case. It's a well known phenomenon. "Signs of Lying: Offering multiple excuses for a situation, instead of just one". One of many sources: https://www.urmc.rochester.edu/encyclopedia/content.aspx?Con...
> All that said, don't hesitate to keep in touch. Your project has a lot of merit, and sounds like it will be real winner with the right team behind it. I'm looking to hearing about your eventual launch!
Anyone who's frequently dealt with proposals and rejections knows that this is a meaningless gesture. E.g. lesser VCs use this "keep in touch" line all the time. After hearing the 10th person say it, who you know couldn't care less about your project, it starts to feel extremely fake.
> Instead, I recommend you talk to one of these local firms:
This part is a nice gesture though. The only part that I'd use from the template.
> I know it's hard to know who to turn to with such an important project, and I appreciate your show of trust.
Perhaps the client had no hesitation at all in sharing his plans with freelancers. Now suddenly you're telling him that he should be concerned and that freelancers as a class are untrustworthy? Even if true, maybe not a good thing to be pointing out to the client.
You grew a steady, low-stress $1MM/year lifestyle business? Great. Now thousands of people are going to see what you're doing, and dozens are going to try to clone you, many of them far better connected than you are. Software is rarely like the e.g. restaurant business, where you can own a scarce piece of property in a good location and don't have to compete with the whole world, or other defendable advantages. Tech startups tend to have far fewer defendable advantages.
Once you realize this, it becomes obvious that "grow or die" is the nature of the business. Not some silly scheme that only exists due to VCs forcing it upon entrepreneurs.
Long-term opportunities that are easily defendable with minimal effort are extremely rare in the tech startups field.
On top of that, guess which type of businesses are, by far, the most attractive to cloners and other competition? - those "ideal" lifestyle businesses that make profits from early on and require minimal upfront investment.
Edit: Or instead of Atari, insert here all the other profitable businesses that you've never heard of, who got crushed after a couple years because they slowed their investments into innovating, assuming that their very steady profits would continue instead of dropping off a cliff. Happened to me twice.
A bit perplexing, but in my experience, VCs who aim lower will often put far more stress and pressure on founders to meet short-term milestones and ramp up revenues immediately. Since the distant billion dollar goal is gone, it becomes all about the short-term.
Aiming for a billion is a great filter too. At the early stages, you desperately need to be associated at least a few investors / founders / employees who absolutely believe you're on the path to something huge. Low goals allows more people to get involved who have extremely misaligned interests, who are often looking for something medium sized to milk, instead of something huge to help grow.
I'd even argue that highly-negative profit, pre-profitability employers can be far better for tech employees.
- High profit = the company's major problems are already figured out, they already have a money-making machine, they don't desperately need you and in fact you may be a liability who could break their already profitable machine. They could fire nearly everyone and possibly continue making bags of cash for a very long time.
- Highly negative in profit = they desperately need someone to solve their problems fast, which in tech product companies is often by making some new tech. Without you, they're nearly guaranteed to end up very screwed.
Wow, it's like they're fully aware of how poor many of their interviewing procedures have been over the years, and instead of making them more employee-friendly to compete with the rest of the industry, they'd rather just punish the employees instead.
Right in line with Google's reputation.
Quite interesting that punctuation co-evolved with musical notations. I had thought that punctuation existed prior to musical notation.
Side note: although American English speakers employ quite diverse pitch and meter patterns, e.g. more homogenous populations, as in many European countries, have extremely consistent spoken pitch and meter usage patterns over phrases and sentences.
This, over-dividing problem, along with the under-dividing problem you mentioned, are both huge hurdles for machine textual understanding and machine translation systems.
On the information-extraction system I'm working on now, roughly 80% of the entities we're trying to extract are multi-word expressions. Very difficult.
[1] https://en.wikipedia.org/wiki/Multiword_expression
[2] http://aclweb.org/aclwiki/index.php?title=Multiword_Expressi...
Yup, that's why whenever I get stuck on the high-level design, I'll often start doing some of the small sub-tasks for which I'm highly-confident will be a part of the end-solution, and for which I'm highly confident that I know the interface they'll need, no matter what high-level approach is taken.
That extra domain work on the sub-tasks often feeds back to helping with the high-level planning too.
This, combined with abstract non-functional high-level prototypes will attack the problem from both the top and bottom. Hopefully you eventually complete the task when you meet somewhere in the middle.
Now try driving in dense NYC traffic, and you'll see humans communicating everywhere. Good luck to self driving cars in that scenario - unless humans are 100% off the roads.
Google / Facebook and many other huge companies are using Theano and Torch7 in production, at scale. The ML industry has been continuously moving in this direction for years now.
On these optimized ML systems, only a tiny fraction of CPU time is spent outside of the GPU. The goal in many of these companies is to migrate all tasks that can be done on GPUs to GPUs, as soon as possible. It's far faster and more cost efficient.
C/C++ is no longer best for speed, not even close.
Now, all that matters is which language has the libraries that make it easiest to get custom code onto the GPU. Python and Lua seem to be winning there, by far.
Also, maybe a dumb question - is there any library or best-practice method for the ensembling of taggers / chunkers? Or must I create it myself from scratch?
Some sentences are just totally ambiguous without context. "Fruit flies like a banana." isn't even good English. Is the sentence trying to say "Some particular fruit flies like a particular banana"? Or "All fruit flies like any banana"?
By the way, Spacy creator - how's the NER coming along?