> could build entire systems at the speed of thought

That's kind of of the problem, you;re limited by the speed of though when coding. For most things I can type waaaay faster than I can think (a faster keyboard wouldn't help, just as neural output interface wouldn't help much either), and information can be displayed on a computer screen waaay faster than I can read it (so a higher bandwith neural input interface wouldn't help either).

What would truly help would need to be much more complex than that, like something that takes input from something, processes it, then spits the output to other part of the brain, hopefully in a way synced with current visual perception. Or takes input from one part of the brain, processes it through some algorithm, then spits output to another one.

My point is that there's a lot to be gained... but I don't think that trivial implementations of BCIs would bring much gain to highly thinking-limited tasks, so the first generation of them would not be of much use to philosophers, software engineers or mathematicians. Probably the first iterations will help most people fighter/drone-pilots or surgeons, who are truly limited by interfaces-bottlenecks, not thinking bottlenecks.

I don't think at the speed of thought means you need to verbalize in your head each bit of your code. You can perhaps visualize it. I can think of a snippet of a for loop a lot quicker than I can make the individual declarations of that loop.

You need to write code, to be read possibly by people without brain implants.

Hmm. That actually opens up a solution: remove the requirement you mention, and we might have sizeable benefits from 1st gen implants in jobs like coding too, we'd just need to switch/translate to languages specifically designed to the make the bes of the implant's particular characteristic.

Though that would soon turn into the mother of all "technical" debt issues...

You type faster than you think? Then who is doing the typing...

I'd interpret it as "once I have planned what I want to type, I can type it faster than I can come up with new things to type." So you always end up pausing occasionally to gather your thoughts between typing.

Are you being ironic? This won't be available to the public, much less even invented for at least a generation or two.

And the developers will have been replaced by AIs by then.

Exactly this. The order in which technological advances will take place as we approach whatever the "singularity" will be will not match our intuition at all!

Imho the future will surprise the shit out of us, despite our ability to anticipate some of the advances... The order shit will happen in will surprise everyone and lead to really unexpected events.

Hopefully we can handle at least some of the things right and steer stuff towards a "successful merge" ( https://medium.com/wordsthatmatter/merge-now-430c6d89d1fe )

It better get sorted out before the oceans turn to acid and everything south of Greenland becomes uninhabitable.

I occasionally look at my editor and know the exact changes or code I’d like to write, and feel a small pang of annoyance that I can visualize it perfectly yet still have to type it out.

Occasionally... unless you're either a genius or you work in highly verbose language and with horrible tooling, you likely spend the other 90% of the time staring at a screen full of code you've already loaded into your brain and thinking stuff like "wtf, dude, wtf, wtf does THAT do, why is that wired to that, how tf does that work, shit, that makes no f sense" :)

So that "miracle" interface would give you, like, what, a 4x increase on 10%, that's like a +4% productivity, in its first iteration, coupled with costs and risks that would make it a really tough sell. We're going to need to go pretty far with the nano stuff to get a good (cost~risk)/(bandwith~latency) out of it. Hopefully there are people working on it and people pouring money into it...

Actually 4x 10% is 40%.

By 4x "increase" in 10%. he means you'd become 4x faster at things that take 10% of your time currently (the other 90% corresponds to the "coming up with solution" / "wtf" moments).

That means for a job that takes 10 + 90 units of time, the 10 section would be done in 2.5, for a total effort of 92.5 instead of 100. That's an improvement of 100/92.5 - 1 = ~8.1% in productivity.

THANKS! I was totally braindead when I wrote that. My 4% is actually the 8.1% you mention, and the way you phrase it makes it much clearer too :)

Suppose you spend 9 hours looking at the code and one hour actually editing code, and the latter is the only part that can be made faster (if we are to believe grandparent). Now, suppose this technology makes you 4x faster at editing code. Then you'll spend 15 minutes instead of one hour editing the code, and a total of 9 hours and 15 minutes. This is roughly 8% less time than you would have otherwise spent. Productivity, being the reciprocal of the time it takes for you to do the task in question, this is an increase in productivity of roughly 8%.

I read it as a 4x speed up of 10% of the work. So what took 10% now takes 2.5%. this leaves 7.5% remaining. Assuming that time is spent doing the same thing then it's 100/(90+0.925*7.5) which is ~3% increase in useful work. I'm sure this math is wrong.

Pretty soon I'll only want politicians and business leaders with brain interfaces.