rabidsnail

fund my new simulated annealing accelerator startup where we etch your model onto an aluminum flake and then hit it with a blowtorch

Housing development is one of the highest on the ripe-for-disruption list, maybe second after higher education. Virtually none of the money these people are going into debt to spend on houses generates real value in their lives, they're doing it because they think they're supposed to. If you can figure out how to build cheap housing on cheap land that offers good living standards (doable now that you don't need access to the power grid) then you'll unlock literally trillions in value.

The only reason to join a startup is if you know the founders and you think working with them would be fun enough that you're willing to give up hundreds of thousands of dollars to do it.

"How is this better than TensorFlow" is the new "How is this better than Hadoop".

I wonder how their distributed training setup compares to mxent ps-lite (in terms of performance and licensing).

I'm probably going to turn this into a blog post but this is the outline of my take on what federation is good for:

* different people have different needs from social networks

* starting new social networks is hard becasue getting a critical mass of users is hard

* people would have social networks that are better tailored to their needs if there was a larger diversity of social networks

* centralized platforms target the average of everyone which doesn't serve anyone very well

* there are two ways to get critical mass: piggy-backing off an existing network or trying to bootstrap one from friends/family/interest groups

* piggy-backing is awkward because the upstart network and the established network both want to kill each other in the long run, but think they can get value out of each other in the short run

* federation can provide the benefits of piggy-backing without being adversarial

* if it's the case that each person has a social network that's right for them that isn't right for that many people, then federation can be a stable equilibrium

Taking time off between jobs is very common and future employers won't think anything of it. The only question is whether you can afford the opportunity cost of not bringing in a salary.

then write some. there's also not that much code here so you could port it to java in a day or two.

Yes, but "cash-flow positive" is the relevant milestone if you're worried about whether the company will be able to stay in business if there's a freeze in the private equity market.

There is no such thing as objectively squicky. I find it squicky.

General rules aren't good for very much other than arguing with people on hacker news.

That's all true, but that doesn't make it less squicky.

Forget abstractions for a second; there are people whom this makes uncomfortable (we know because they said so), and the cost of that is much higher than the very small benefit of using one image over another as a test image. Simple cost/benefit analysis says to use something else.

Also if you don't know what your target palette is you can do k-means clustering over colorspace, and the palette is the cluster centers (each point you're giving to k-means is an <r,g,b> or <h,s,v> vector).

For small distances, yes. If you plot a 2d projection of a dataset that doesn't have much structure you're going to be reading patterns into whitenoise (though this data has some pretty clear clusters, which are probably real). If I were doing something other than writing a fun blog post I would have done cluster analysis with something like DBSCAN.

There's no UI for if there are no matches; it just does nothing. Try searching for \.com or something.

Edit: I patched it so it displays an alert if there are no matches.

Non-spam bots generally don't follow each other or link to external websites. (I'm also the author of one of the more popular image bots https://twitter.com/a_quilt_bot)

(original author here) it's a spambot network. If you click the link in that post to the interactive version (this: https://pile-of-junk.s3.amazonaws.com/twitter_scatter_10k.ht...) you can see for yourself.