When I was a rails developer, the ‘done thing’ was to simply throw hardware at issues like these as an acceptable tradeoff for productivity. If you cared about this sort of thing, you’d use something more formal. I find it personally difficult to calm my pearl-clutching perfectionist tendencies to embrace that approach but I can’t deny it does work :)
The article says "Augment is using fine-tuned “industry-leading” open models of some sort" - correct me if i'm wrong (happens often! :) ) - but does this mean that they're using some open model from huggingface, wrapping some UI around it, and voila they're worth a billion dollars? Hmm.
I have an M1 ultra, it compiles large Xcode projects within a few percentages faster than a colleagues M1 Max; next time I’ll simply buy a max and save the money. This might be more an indictment on Xcode however. It seems a marginal business for Apple to bother making these, I wonder how they justify it.
I wonder... perhaps linux driver issues were more entertaining 20-30 years ago because there were less peripherals to worry about. Eg no wifi, no power management, etc etc. Nowadays computers are more complex, it's more involved, more exhausting?
I know i rarely read the articles, i just read the headlines and the small amount of blurb accompanying it. Perhaps a lot of people are like me, and the newspapers are supplying the headlines/blurbs/image and not even getting a clickthrough, and then people don't feel the need to visit the news sites to get 'up to date'. Perhaps that's why they feel they're missing out.
Oh, many years ago I bought into the whole 'write a blog, collect subscribers, create a training product and sell it to them, retire on the beach' idea and gave it a go. I now think that's exceedingly unlikely, of course. Sorry you find it annoying, i'll have to get around to removing it sometime.
Recently tasked with implementing a Kalman filter, I found it very very difficult to find good resources that explained it in language that made sense to a developer like me. So after spending a month learning it, I wrote a couple posts on it, perhaps someone might find it helpful?
As a developer I found the maths made sense only after implementing it, ironically. I guess we learn by building on top of what we already know? Is there a term for that?
In my experience the vacuum holds the bottom half immovably in place, and I have to pry it out. It looks nice in that gif but in practice I think they need to test a little more. Perhaps it’s a humidity thing?
One interesting thing to me is that apple has now put a price tag on their payments processing: 3% (being the discount of this 27% vs the normal 30%). Their payments are pretty convenient, i'm surprised they don't consider it worth eg 4-5%.
So they're basically saying that their SDKs (+ distribution) are worth 27% of sales.
I wonder if one could make the argument in court that if you don't use their SDK, eg you use react native or flutter, you shouldn't pay 27%. Yes, i know those frameworks still use apple's SDKs, but they commoditise them such that you might make the argument these SDKs aren't worth any more than any other sdk such as android.
(I'm skipping distribution in my argument too, for simplicity's sake)
I’ve heard it said that the existence of objective evil is only explained by the existence of objective good. And yeah… some of what happens in the world just has to be a step beyond subjective evil. Please stay curious :)
Edits to google maps typically get reverted after a month, at least they do for me and others I’ve read about online. I worry they’re resting on their laurels
Notice that so many parts are metal and only needed sandblasting. Modern German cars use plastic intake manifolds and rocker covers, they’re all cactus before it’s 10 years old.
I like to append ", unless I'm missing something?" in a similar manner, might be useful in situations where 'made a mistake' doesn't actually make sense :)
Thanks a lot! That does explain some of the ideas behind it nicely. Sometimes when looking at equations it's easy to miss the forest for the trees so to speak, so this helps.
I've been asked by a friend recently to implement a Kalman filter for a side project. Among many others, I've read this website and still have no idea how to implement it. They all seem to be in the style of 'draw the rest of the owl'.
Can anyone point to anything that explains it in a manner a programmer would explain (eg array iteration instead of sigma notation) ?
As far as i understand, Kalman Filters seem to be a moving average of position, velocity, and (perhaps) acceleration, and use those 3 values to estimate the 'real value' as opposed to what the sensor is telling you.
Amazing how capable tools can be, in industries that simply cannot get their work done without them and are willing to pay a healthy amount. Sometimes I wonder how good our dev tooling would be if we were willing to pay thousands a year, as opposed to the time-wasting bug-fest that is Xcode. Not singling out Apple, I’m sure all are comparably bad, it’s just what I happen to use :)
I'm a little concerned that the downturn in that graph coincides with changing from measuring one financial instrument to another. But i dont know enough about those instruments to know if that's an issue or not :)
That sounds really exciting! Do you happen to know if such boards will be competitive with the RPi5 eg quad-core etc. I love the idea of RISC-V resolving some of the drivers dilemma that SBCs seem to face. I recall some talk of extensions that might make it viable as a GPU core one day. Not to rain on the RPi5 parade of course, i'm sure it's a fantastic bit of kit :)
Thanks! I take it that this means a Generalised Linear Model? Could i ask for a link to a relevant article to get me started on the flavour of GLM that you recommend?
On this topic, does anyone know of a suitable time-series forecaster for multivariate analysis? Eg 8 independent/input variables, and one output variable? I've been using multiple linear regression (which works impressively!) but it doesn't take into account the time series, only the single prior day of inputs. Thanks :)