Perhaps instead they could build a central store of potential ad clicks and their base rate likelihood of being clicked. Then, instead of clicking everything, they could click ads at the same rates an average user would, across the network.
I'd learned Dvorak using a Kinesis keyboard. I use both Dvorak with the Kinesis and QWERTY with a standard keyboard. The slight difference in profile is enough that I've never had interference, and use both comfortably. That said, it hasn't been an enormous speed increase; I'm roughly equally as quick with either.
The author didn't do a very good job interpreting these statistics. The conditions that would exist when someone would choose to invest in solar panels exclude a large fraction of homes; solar panels are a recent enough addition for most that homes which have been allowed to fall into disrepair are excluded. They further exclude homes that don't have a view, homes that are part of a multi-unit bloc, and, in some cases, duplexes.
Their success metric may just be reversion to the mean; since they started by selecting the people who had made the most calls, you would expect the number of calls to come from those specific people to fall in the coming years. Since there was also likely some underlying health condition, you would expect the between-year variance to be large. It might well be that their program has no effect whatsoever.
If no other group is calling at a similar level, then it's not reversion to the mean. The mean has actually been shifted downwards.
> Since there was also likely some underlying health condition, you would expect the between-year variance to be large. It might well be that their program has no effect whatsoever.
The rate dropped significantly each year from 2015 through 2017 -- 3 years in row. It's possible this is pure chance, but it seems rather unlikely.
If they're really tracking the same 100 identified people, with no replacement, then you'd expect the same result even without regression to the mean, simply because very sick people tend to die, and dead people tend not to call 911.
One would hope that they're not doing that, and it's just that the article is imperfectly worded.
With the cost of re-training for new jobs going up, and the value of un-skilled labor going down, there are certainly more and more folks for whom it's more expensive to get a job than any benefit that they'll receive from it.
So there can be lots and lots of stuff, but if you aren't very good at learning, and you don't own capital, the world can become quite unfriendly - even if we've gotten really good at making things and most of us own quite a few digital watches.
Pikimal is looking for people interested in NLP, search, and information retrieval. We've a fast-growing collection of facts, facets, and evolving definitions of common adjectives, but we'd like some help making these assets more useful.
We code in Ruby, but have had good experiences hiring people without Ruby experience. We're also looking for a best-in-class SQL tuner for contract work.
Pittsburgh, PA (no remote) http://pikimal.com/jobs
We're looking for Semantic Web and Ruby Developers but if you're a strong developer who doesn't know Ruby yet that's no obstacle. We have extremely flexible hours, collaborative coder DNA, good tools, a strong team to work with, and great health care.
Pikimal is working to change how people use the web to make decisions. Once users tell us what's important to them, we can tell them what's best for them. Since all of our recommendations are based solely on facts, users receive results separate from marketing.
Please include a link to public code you've written or your Github repo when you apply. Feel free to reach out directly to my first name @pikimal.com
There are other options besides social or text-based algorithms. My company, Pikimal, is pursuing one - we're pulling together facts on items and are allowing people to weight those facts to dynamically create recommendations. We've added some "user rating" facets but are just barely past our minimal viable product.
I very much agree with you that social is limited. It's a filter which avoids spammers, but it also filters out experts and users. It also solves some of the problems introduced by the generic nature of search algorithms.
I'm betting my time on the idea that the solution which wins will combine an understanding of the product space, the value of new features, a current understanding of price, and which can be customized transparently to the needs of the user.
