If the quality of the software in his cars is anything to go by... don't think I'd be keen on putting one of his devices in my brain

Not to mention the telemetry those cars phone home with, I certainly wouldn't want that in my brain either.

I put this in the same category as drone deliveries, except a billion times more complex.

Agreed, not gonna happen.

But it is such a fine and deep money pit. It is a great bubble of wishful thinking. You can burn and destroy billions and billions before anybody will pull the plug. Think of all the possibilities to redistribute wealth to the deserving rich.

"I think in the future you will be able to save and replay memories..." Doesn't say Neuralink can do it, or even that they ever will.

The "I think" part tells us it's just Musk's musings.

"in the future" acknowledges we aren't there yet.

The rest is just the description of the problem.

Calculating the data required to hold the entire brain state seems like saying there's X number of photons in this scene so we can't possibly record a video of it. At present we just don't know how much data is required to hold a memory, or how to capture and replay it.

I can't imagine we'll have this technology soon, and maybe never, but wouldn't rule it out.

"Tim Verstynen, associate professor of cognitive neuroscience at Carnegie Mellon University, took to Twitter to post a "back of the envelope calculation" to explain why exactly Musk's claims were so far-fetched."

Confucius he say; 'Man who say task is impossible, should not get in way of man doing it.' Good on Elon for refusing to settle for the status quo, and daring to try and make the future happen.

Why is it that Musk’s enterprises seem to have one of two aims? Tax dodge or government handout

I wonder which one this is?

I'm a cognitive neuropsychologist, but with a bit of a computational background (spent a summer at the Santa Fe Institute, refereed things about physics of computation, that sort of thing). Dr. Verstynen is right to draw an analogy with chaotic dynamics, but the brain situation is even worse. Typically, when one studies chaos the system is isolated - in brains new perturbations from the outside world keep bumping into it and the system never has the chance to settle into stable attractor dynamics - it is always on the way there - it never arrives (until we die). There are probably transient, sort of predictable, quasi-stable states induced by signals from outside world, but I don't think we have much of an idea at all about how these are formed and how they affect the rest of the brain. All a bit hand-wavy, but, at this stage that's probably all you're going to get. Elon is talking out of his hat.

Yrs truly, A Professor.

Would you go on record and say this will never happen?

Going by the downvote I'd say no, you won't go on record. In which case, you and Elon are saying essentially the same thing; we're not anywhere near yet, it's phenomenally difficult and unlikely to appear any time soon, but let's take the first steps and see what happens.

"Absolute balls. This is god-level cockwaffle, typical of the manballs beloved of these ignorant dorks."

I don't have Twitter, but I may have to sign up only to follow this gentleman! Bravo! Bravo, I say! Brought a tear of joy to my eye, it did.

Will he also be able to analyse all sorts of diseases from a few drops of blood?

For all the people who are going to cheer this sort of stuff on, because, like, Elon's a megarichtechgenius, I would first recommend reading some of J.G. Ballard's short story books. They are full of horrendous ideas like this, with the end result often being that the human guinea pigs they're tried on are left permanently screaming in terror at the irreversible hell they've ended up in.

Also, they're much less time consuming than spending an hour watching Black Mirror tediously labour its "bad tech use" point.

Modern research suggests strongly that he brain stores memories by making new synapse connections. These connections fade over time and eventually break. Whenever we recall something (perhaps only by dreaming about it), the strength of those connections is reinforced. So memory is a dynamic thing, wholly bound into the dynamic evolution of the brain.

To artificially recall a memory you would have to identify and artificially stimulate every synapse involved in that memory. This demands two particular technologies:

1. Wiring up every synapse in the brain, and updating that wiring dynamically in real-time.

2. Identifying each individual memorable event and the synapses associated with it, and storing that association.

So you need a database index as comparably vast as the raw brain configuration itself, as much hardware wiring stuffed into in the brain as there is wetware wiring, and a database technology sufficiently powerful to operate the whole shebang in real-time.

That data system will effectively be a brain in its own right. It is more likely to sever the connection and sign on to Facebook than it is to waste its life backing up Musk's distorted reality.

Idiocy comes in many forms. Not stopping to reality-check one's ideas is one of them. It has been said that genius is an infinite capacity for taking pains. Poor Elon risks becoming a classic example of what happens when you do not stop to do so.

The professor knows his field. However, he doesn't seem to grok Musk's usual MO - i.e state an improbably far off goal as a definition of a direction, then build a profitable business model that can does R&D in that direction.

The professor is right to take every opportunity to remind us of how inappropriate the computer analogy of the human brain is. It is an analogy that has been too often used in popular culture. Musk already knows this. Neuralink could a way of learning more about how the brain does process information.

Why? Because the short term market for devices that can treat range of medical conditions is large and clear. People have greater chance of getting neurological disorder in their lifetime now that we don't smoke ourselves into early heart attacks. Traumatic brain injuries, spinal injuries and amputations have risen due to military activities, and the medical techniques used to treat them have already been improved due to medics' greater experience. We already treat some Parkinsons patients by placing electrodes in their spine.

Saying that you're walking to the sun after lunch is just another way of saying you're walking West.

It should also be noted that Neuralink's demos have been to attract potential employees (who one assumes know their field and know that 'uploading memories' is decades away even if it ever is possible), not to attract funding (where one might be more careful of one's language because fund managers aren't usually technical specialists).

Remembering and learning how different our brains are from computers is actually a fun exercise (well, fun for the sort of folk who read the Reg, I hope!), and might give clues to how computers might be improved. An example might be the layers of processing that start at our eyes, with information being discarded at each stage. Our brain never receives a high def video recording (which would be the crude video camera computer analogy). Now we have Sony building patten recognition algorithms into the camera sensors so that some processing can be done without reduntantly shunting HD video around networks (waste of limited power and bandwidth).

It could be that in the future computers will become a more appropriate analogue of brains, because we will have created computers that are more like brains. Or rather bodies, since the human brain is lazy (er, efficient) and will outsource processing and memory to our nerves, our guts, our bodies, what we're holding in our hand at this moment, what room we're in, the colour temperature of ambient light etc

There's only one way to find out. And dissecting twitter comments ain't it.

Similar misconceptions underpinned the 'beaming down' in Star Trek. Quite apart from the obvious defect of there being in many instances no receiving apparatus to do the re-assembly (as when they beam down onto a planetary surface in the open), the humungous amount of information transfer required to reconstitute a human plus all the extras (clothes, weapons &c.) and reconstitute them accurately precludes feasibility - if for no other reason than error rate.

It also falls down over the ambiguity of whether the process transfers the original matter or replicates it. There's an episode in series one ('The Enemy Within') that suggests the latter, as Kirk gets split in two - a 'good guy' and a 'bad guy'. They're both Kirk's original height after the split, so their combined matter volume is twice that of Kirk before the split. If beaming was matter transference and the split was 50/50, we'd expect to have two rather smaller Kirks, each about 80% shorter than his 5'9" (i.e. about 4'7" each).

The problem with the replication method is that once you've beamed down it's not really you any more.