Posts by Spamfast

The Smoot – How an MIT prank became a lasting unit of measurement

Re: the thickness of screw threads was not fully standardized in the US

Whereas the thickness of the electorate clearly is.

(Unfortunately in the UK as well.)

Queue the "Americans will measure with anything but metric" joke lol.

Cue the "Americans will measure with anything but metric" joke. LOL

To cue is to do something that allows or provokes a follow-up action. To queue is to do what the Brits spend half their time doiing.

Nearly 3 out of 4 Oracle Java users say they've been audited in the past 3 years

Just say no.

These days, one should be fired for choosing Oracle.

Or Java.

Remembering when NASA stuck a Space Shuttle on top of a Boeing 747

For one due to lack of a place to put it

Build it, have it clutter up the dining room table for a couple of weeks until it starts collecting dust and complaints from the significant other then break it up and put the bits in the big box under the bed.

Or are you one of these sadsacks who keep all their built Lego kits in a cabinet for ever - Airfix/Revell Models for Dummies style?

Do you superglue the bricks together to stop your creations from falling apart as well?

¡CHECK THE ICON! ;-)

£127M wasted on failed UK nuclear cleanup plan

THE NLF is window dressing at best.

It only covers specific existing facilities.

Its funding doesn't even come close to covering the real costs of decommissioning those, let alone any new ones that might be built.

It is also largely controlled by the nuclear industry itself, in most part Electricité de France. Hardly an impartial overwatch.

I strongly suspect that 'the current form' is a lot better than the 'currently being decommissioned form'

I do hope so but given the nuclear industry's track record I find that extremely doubtful.

Remember, the shareholders don't pay for the decommissioning. By that time the company has been dissolved and the taxpayer picks up that bill for however many decades or centuries it takes.

Forked-off Xlibre tells Wayland display protocol to DEI in a fire

Re: Code talks

We agreed then and there to use 'primary' and 'secondary' instead for the project we were working on.

Replacing 'master' with 'main' or 'primary' or 'trunk' in repo branches is fine. (Although of course the word 'master' has many uses other than master/slave - master craftsman, master key, etc.)

Using primary/secondary is not appropriate for some situaton e.g. where one entity (software module, program, device, etc.) tells another what to do. In that case, if master/slave is deemed offensive which I can understand given the continuing inequities from the African slave trade, some choices might be controller/controllee, owner/executive, director/worker etc. Primary/secondary sounds like one is a backup of the same function for when the other fails.

Perplexity offers training wheels for building AI agents

Surely if you fit them with nappies it'll block their output entirely?

Not that that is a bad thing per se.

Yet more Merkinization I see.

In the UK they're called "stabilizers" not "training wheels" - at least they were when I were a lad. (Flat caps, licking road clean wit' tougue etc, etc.)

(Icon for my comment, not the article!)

Yolk's on you – eggs break less when they land sideways

Re: Lilliput and Blefuscu

Ah, I see.

(Icon for my own obtuseness.)

Lilliput and Blefuscu

The paper's authors have referenced Dean Swift's seminal work on taxonomy of egg morphology but have seen fit to ignore his universal if not uncontroversial nomenclature big endian and little endian for blunt end and sharp end† respectively.

And here's me thinking it was Jonathan Swift. I wonder if they were related?

Apple slams door on Fortnite's stateside iOS comeback

Re: To hell with spent fruit

Actually, the various versions of the Christian Bible and the Hebrew texts upon which they are based do not specify the type of fruit that grows from the tree of the knowledge of good & evil.

So since God is omnipotent, it could have been a genus of the phallic banana or even the smelly durian.

Raspberry Pi, especially now it's gone public, might even be an instrument of the evil one since it's an aggregate fruit. ☺

(But I'm also an athiest & I agree that Apple is a nasty corporation, although there are very few that aren't.)

37signals is completing its on-prem move, deleting its AWS account to save millions

Denzel DIY calling

I run my primary server (SMTP, IMAP4, DNS, HTTPS media/SVN/Git/Web/reverse-proxy) on prem via VDSL. All data backed up every night to 7-day, 5-week & 12-month incremental rolling via VPN every night offsite to a server at my secondary site and pay Oracle the outrageous sum of ~£1.90 a month for two additional VMs running DNS resolvers & SMTP MTAs in London & Frankfurt with the added bonus of a private VPN (that isn't blocked by the providers like most subscription VPNs are) to access German- & EU-only services via the 1TB/month allowance.

Of course I'm not exactly in the 200 petabyte range - more 10 terabytes.

As the icon says, "I'll get my...".

More seriously, cloud & other managed IT is great for start-ups and for companies whose primary business isn't IT-related but for large companies whose whole service is based upon large amounts of data and/or compute it's pretty obvious that on-prem - perhaps with some co-lo data centre - is going to be far more economical since you're already going to have staff with the requisite skill set.

Ghost in the shell script: Boffins reckon they can catch bugs before programs run

Re: Two easy bash script tests

Moreover set -e is overrated and unreliable. If a command can fail the script should check for the failure explicitly.

I agree that if the failure of a command would lead to the script leaving things in an inconsistent state, then it should be explicitly invoked via if or by checking $?.

But checking the result of every external command, shell built-in & shell function that might fail is exhausting if a simple abort of the script - maybe with a trap EXIT to clean up - is less error prone and in any case set -e will act as a last resort assertion.

Re: Two easy bash script tests

I know it works and is the more portable thing to do but #!/usr/bin/env bash seems illogical from a purely abstract point of view.

Avoiding an absolute path for invoking one program via the absolute path to another seems daft!

Anyway, back on topic, some other ways of improving code quality is always double-quoting substitutions (completely e.g. cmd1 "$(cmd2 "${var}")"), never using backticks for command substitutions, using xargs or mapfile instead of command substitution.

As someone said, installing shellcheck to lint all your scripts is a good place to start and will pick up these and many other potential gotchas.

Re: Two easy bash script tests

I agree that the exit-on-error & error-on-undefined flags should always be set either via the shebang or the set command at the top of the file.

If using bash, then set -o pipefail is also a good idea.

But the first line should be #!/bin/sh -e not #!/bin/bash -e unless the script is actually using bash features like shopt -s lastpipe, arrays, local variables etc.

Using bash for shell scripts that only need basic POSIX shell features wastes a lot of resources, especially if said script is going to be run repeatedly from cron or via other background triggering.

On 64-bit Raspbian for example:-

$ ls -l /bin/{ba,da,}sh

-rwxr-xr-x 1 root root 1346480 Mar 29 2024 /bin/bash

-rwxr-xr-x 1 root root 133640 Jan 5 2023 /bin/dash

lrwxrwxrwx 1 root root 4 Jan 5 2023 /bin/sh -> dash

Google goes cold on Europe: Stops making smart thermostats for continental conditions

Re: Again

My step-daughter was 'given' a Nest thermostat by her electricity company that steadfastly refused to connect to her, her mum's or my WiFi - after which I discovered it was a known issue with no fix other than to upgrade to a newer model which was basically 'buy a newer model'.

Matter's definitely a step in the right direction but things like ZHA & ESPHome already provide open standards that are more capable.

I dipped my toe into home automation with TP-Link Tapo just using their app. Now anything I buy has to have a local (i.e. no manufacturer's cloud server) integration for Home Assistant and/or Frigate. Thankfully Tapo gear is in that category as is Sonoff, SwitchBot, Govee, lots of BLE gear, most Zigbee (ZHA, Zigbee2MQTT), Matter, Thread, anything that comes with or can be converted to ESPHome and quite a lot of other stuff. You can then integrate the lot, automate it however you like (without giving ITTT the keys to the kingdom) and access it from anywhere using the HA app via your own HA server with no third party peering over your shoulder. You can even connect it to Alexa or Google smart speakers for voice control if you feel like throwing away the privacy you've claimed back but there are local HA smart speakers too with open source firmware. HA will run on an RPi or low end fanless PC or as a VM or Docker image on your existing server so chews bugger all juice.

Satellite phone tech coming to your mobe this year – but who pays for it?

Re: This is What'll Happen

Oh 'eck. Two pence piece payphones. Push button A to connect, eh?

I used to design some of the original digital electronic on-the-roll/freefall coin validators (and the payphones into which they were fitted) but I still love the old electro-mechanical ones.

I also think Strowger subscriber trunk dialling racks were way cooler than System X and similar all electronic stuff.

They just sounded way cooler!

(See the icon.)

Re: This is What'll Happen

why keep up the expensive hardware on the ground when you've got expensive hardware in the sky

Because the hardware on the ground can provide way more bandwidth per buck than that in the sky. It makes much more economic sense for SkyNet (I'll be back) to provide the fill-in for blackspots rather than having it handle the whole load. It would be very expensive to provide the entire mobile bandwidth from LEO, not to mention resulting in so many satellites buzzing around that it would end up like a neutron cascade if any of them got clipped. (Think 'Gravity' but a realistic version.)

I could run my entire network connection via an LTE/5G modem but it's much cheaper to have a FTTP or VDSL connection and WiFi at home and just use radio when on the move. Similar physics.

Nuclear center must replace roof on 70-year-old lab so it can process radioactive waste

I fear the battle has already been lost.

A meter is a device that measures something whereas a metre is a unit of length.

On the other hand, as the Moties might say twice, the Germanic languages (German, Dutch, Danish etc.) from which English originally derived use Meter/meter. The 're' instead of 'er' is a Romance influence from the 18th & 19th century gentrification of English as is 'ise' instead of 'ize'.

Re: "Confident they have an alternative"

Of course they do.

As with other undesirables, they can simply bribe someone and ship it all to Rwanda.

Re: Alternative option.

Don't go building new nuclear power stations until the end-process has been agreed.

There's the rub. The first job when considering whether to go ahead with a project needs to be to calculate the whole life-cycle cost of building, running & decommissionng the facility and the total cost of safe processing, storage & disposal of the residues for however long is required afterwards.

Of course if that were ever done properly, nobody in their right mind would ever underwrite a new nuclear plant.

I don't have a fundamental problem with nuclear power - although I wish there was more research on thorium tech instead of uranium - but the fact is fission is, always has been and probably always will be the most expensive grid electricity source we have. In the UK at least, a nuclear power station also takes thirty years from breaking ground - never mind finance & regulatory stuff beforehand - before it's generating.

It would be far lower financial risk and faster economic & environmental (the two are the same really) reward if the money were spent on storage tech - thermal, potential energy or whatever - and renewable generation.

NDA

controlled by the NDA

NDA?! Shh! You're not allowed to tell anyone!

Bill Gates unearths Microsoft's ancient code like a proud nerd dad

Re: The Moral of the Story

Worked for Marion Morrison too by all accounts. "You can ride a horse, yes?" "Sure."

Re: Lotsa lookups

or have terrible comments

i++; // Increment the counter.

Windows 11 adds auto-recovery, kills offline setup loophole

Re: Meh

Do you have <firsname>@hotmail.com ? They're the real elders.

Not quite, but I do have [e-mail] - actual, real and quite common first & last names at that.

I also have the same firstnamelastname as my separate Skype ID from its pre-Microsoft days.

Not that I use either much anymore - there are better voice/videocall services out there and I have my own email domains & servers.

UK govt data people not 'technical,' says ex-Downing St data science head

We could do with some more actual STEM experience in our elected officials as well.

Back in the mists of time I won a Manchester Lit & Phil computing prize and was congratuled by the then Minster for Industry & Information Technology Kenneth Baker. I could tell then that my 18 years of life experience had taught me more about industry, science & tech than his fifty years of paid-for history & law degrees and 'accepting campaign contributions'.

We still have equally talentless twonks pushing for the latest flavour of the month that the government service proivders like Capita and Fujitsu are paying them to promote.

Plus ça change, plus c'est la même chose.

Hm, why are so many DrayTek routers stuck in a bootloop?

Re: Inscrutably bad router

Zen here too at two SOHO residential sites and I got FRITZ!Boxes too with built-in FSX & DECT telephony. Only use them for telephony & the VDSL modems - use OpenWrt for everything else.

FRITZ!Box are way better than what you'll get from the main ISPs though and they do get upgrades from AVM from time to time.

I wouldn't buy Netgear, D-Link or Cisco routers or access points ever, except for comsumer use where I know they can be converted to open source. (Netgear WAC124s used to convert to surprisingly good gateway routers for ~ 50Mbps DSL connections and they cost diddly-squat.)

If one must choose proprietary, DrayTek are okay. But there are several 'industrial grade' hardware manufacturers that support open source projects like OpenWrt directly.

Oh, I wouldn't buy a DrayTek to run OpenWrt.

Try a GL.iNET Flint. One hundred & sixteen quid gets you a 64-bit quad-core ARMv8, 1GB RAM, 8GB flash, two 2.5Gbps & four 1gbps Ethernet ports, separate 2.4GHz & 5GHz WiFi phys, USB3, running latest OpenWrt or GL.iNet's front-end on top if you prefer. Linux DSA architecture but you can let the 5-port switch do all the VLAN switching work itself as well and NAT & flow offloading to the WiFi & Ethernet hardware means the CPU is laregely free for other uses.

What exactly, can your proprietary router do that this cannot? And how much do you have to pay for that propietary router - at least five times more than I did for similar spec & functionality.

When just running as a router, my CPU usage barely gets out of the sub-5% range. So I can run services like NFS, Samba, Apache, Postfix, Dovecot or Asterisk natively or there are plenty of Docker images that run fine too.

Fully open source including the build environment so I can tweak & build the firmware myself if I want to or just use the web builder or the package manager post installation. And no vendor lock-in.

Problem: proprietary firmware

Solution: OpenWrt

UK's first permanent facial recognition cameras installed in South London

Re: Success rate?

How long before the coppers get fed up with wasting their time over all the false positives and quietly stop using it?

You're trying to get Met coppers to get on their feet and start patrolling the neighbourhood in the way that has been shown to have the greatest effect? If you can't prise them out of their patrol cars you certainly won't get them to come out of the nice warm surveillance suite with its coffee machine. Except of course for a profitable bit of overtime on a demo somewhere or to get brownie points for arresting Mr Winston Kodogo for walking on the cracks in the pavement or for wearing a loud shirt in a built-up area after the hours of darkness.

Not a lot seems to have changed over the past forty years, does it? https://www.youtube.com/watch?v=teSPN8sVbFU (Is this your hedgehog, sir?)

Museum digs up Digital Equipment Corporation's dusty digital equipment

Re: "I agree about the 6502 'running rings around everything.'"

a prioritised address generator to the fixed IRQ indirect address

Very cool. Your own NVIC but even faster! Kudos.

I quite like PICs too though but the 6502 was definitely easier - not a big fan of Harvard Architecture. Don't mind it in things like ARM IBUS/DBUS if it's hidden below the memory addressing but it just makes life difficult if it's visible to the programming model. Same way I don't like the separate memory and periphiral access model in the Z80 & x86. (Whay have IN/OUT instructions when MOV/LDR/STR work just as well!?)

I agree about the 6502 'running rings around everything.'

Back when I & a friend - both of us sixth formers at the time - were setting up the computer science department for my grammar school so it could start offering computer science O-level to the lower years we started off with one ITT 2020 Apple ][ clone & one Tandy (Radio Shack) TRS-80, both with cassette tape recorders for mass storage, by the way. On paper the Z80 instruction set looked far more powerful but in practice the TRS-80's Z80 was seriously outperformed by the 2020's 6502 in terms of clock cycles required to perform a counting loop and similar basic operations. On the other hand, the TRS-80 didn't pop its RAM chips out of their sockets in warm weather but that's a separate issue!

The only thing from the Z80 that I wished the 6502 had had was a 16-bit stack pointer. The 8-bit SP and fixed stack page at 0x0100 basically meant higher level languages always had to implement a call/parameter stack in software. Having said that, it didn't slow down the amazingly fast BBC BASIC interpretter on the BBC Micro which had named & parameterised functions & procedures. This was several times faster than the BASIC on the expensive pre-PC desktop HP machines I later used at college. The BCPL, LISP & Prolog ROMs were pretty nippy too as was Acorn's ViewSheet.

Ah, nostalgia is not what it used to be!

'Dead simple' hijacking hole in Apache Tomcat 'now actively exploited in the wild'

Yet another reason not to use Java-based horrors like Tomcat or JBoss for server backend stuff - if the ludicrous compute power requirements weren't already enough.

Cheap 'n' simple sign trickery will bamboozle self-driving cars, fresh research claims

tested on a "rooftop parking structure"

The Weathergirls might have to update their lyrics. "It's raining Benz, Hallelujah."

Does terrible code drive you mad? Wait until you see what it does to OpenAI's GPT-4o

Re: Insidious bias

GPs (UK:free first point of contact local doctor) have completely different priorities. Typically making the best of limited resources to do the best they can. (Define 'best'! Prioritise Aunt Ada's cough or Baby Brutus' rash?

Anyone who thinks that this is the priority of the doctors who own GP practices is as deluded as those who think LLMs are GAI and clearly hasn't been observing the evidence of the past twenty years. The amount of taxpayer money given to these privately owned businesses has skyrocketed and is about to go up again both in absolute terms and amount per registered NHS patient. Remuneration of the owning partners has jumped massively. At the same time level of service, continuity of care, patient satisfaction and the hours worked by the owning partners has nosedived.

When the NHS was introduced, GPs managed to maintain their for-profit status as a quid pro quo for supporting its introduction. In the 21st century they've discovered they can use the NHS as a cash cow without actually providing any accountability or improved service for those who provide the money.

What annoys me most is that whenever I go into my local GP-owned health centre, it's never busy. Yet they claim they need more and more of the NHS's money because they're overrun with increased demand. There's some very creative accountancy going on somewhere.

Early mornings, late evenings, weekends. Useless users always demand support

Re: Dual SIM phone

work to live, dont live to work

Haleluljah! Have a beer. And while we're propping the bar up, let's not bang on about work.

Unless they're paying you at least time an a half for being so - not for when you're actually called - then on-call is a scam to avoid employing enough staff.

For those still sufferering, next time you consider changing jobs then buy a throw-away SIM and use the number for the agencies (spit!) and your new employer. Stick it in a $10 burner phone and you can shred it later.

UK employment contracts often have weasel words such as "your hours are 37.5 per week ... but you may be called upon to work beyond those in exceptional circumstances". This is unenforcable because they don't define those precisely and contract law favours the one who didn't draft the contract.

'Maybe the problem is you' ... Linus Torvalds wades into Linux kernel Rust driver drama

Lilliput, Blefuscu & boiled eggs

If you hand the keys to the kernel level DMA API to any programming language then the code that uses that API can not be memory safe.

use of unsafe in Rust

Indeed.

I don't do a lot of application (i.e. program launched in userland on top an OS) programming these days but C would not be my first choice there anymore - I'd go for C++ making sure to avoid the use of memory-direct constructs. Rust looks like a good alternative in this domain. To be honest, when I have to knock something up to support what I do Python is often the better choice. My preferred arena is bare metal - sometimes RTOS mediated - design but I've also done work in Linux kernel space device driver stuff. The QNX micro-kernel userland driver model was a hell of a lot easier but Linux is okay once you understand the quirks of the subsystem in question.

Anyway kernel programming, especially device driver development, is inherently memory unsafe, regardless of the language used. If you have to and are allowed to bugger about with the contents of the physical memory map, device registers, interrupts and the rest then you better get it right and it doesn't matter what type of runes you're using to do that.

Rust may be 'safer' than C but at the end of the day it's the discipline in coding that counts and frankly, coming from safety critical, a lot of what I see in the Linux kernel's C code is god-awful in design, implementation, efficiency, documentation and error confinement. If this is the culture then I doubt Rust is going to make any difference and just adds more chances of screwing things up since there are now three times as many things to go wrong - the C, the Rust & the bindings between the two.

M'eh! Like I say, I prefer bare metal anyway.

Canvassing apps used by UK political parties riddled with privacy, security issues

Re: Flogging a dead horse

Dear Marjorie Proops, I can only achieve gratification via sadomasochistic, necrophiliac zoophillia. Is this okay or am I just flogging a dead horse?

(see the icon)

Re: Quote: This will further undermine public trust.

I upvoted you because I agreed.

I think it has already hit rock bottom.

But the events of the past few years have shown me that the current geological formation has an infinite depth.

Trump admin's purge of US cyber advisory boards was 'foolish,' says ex-Navy admiral

Re: Bad timing ...

kiss the one he sits on

In that case ...

Bender for president!

Re: Is 'learnings' a word?

You don't have to be smart to get into Harvard or Yale - or Oxford or Cambridge. Having rich parents generally makes you a shoe-in, especially if they themselves are alumni, having got in by the same method. Once there, you are almost guaranteed to achieve a degree in some fatuous subject like PPE or law without having to do any real thinking. It has to be this way if we're not going to run out of "qualified" senators & MPs, of course.

Only poor people have to be smart to have a chance at elite universities, and they also have to take on a lot of debt to do it.

Interesting side comment, despite its repeated claims to be the land of opportunity, the USA is one of the least socially mobile countries in the western world. Even the UK has better statistics. (Look 'em up yourself by the way. This is a rant not a scientific paper!)

WFH with privacy? 85% of Brit bosses snoop on staff

Re: so management are not measuring output

And there lies the rub.

Productivity is, by definition, getting the product out in a profitable way, whether that be physical widgets or whatever. As long as that happens, management shouldn't care about how the ones generating the product are doing it provided that they are happy.

Unfortunately, management defines its own productivity by how much it is seen to be interfering with the producers' activities.

IT department heads suffer from the same insecurity.

It's about justifying one's own existence.

Re: Responsible adults

If you don't trust your employee, then why did you hire him in the first place?

The corollary, of course, is that if your manager insists on surveillance then he doesn't trust you and for your own career & psychological benefit you should find another employer as soon as possible.

Tiny Linux kernel tweak could cut datacenter power use by 30%, boffins say

Re: #pragma pack(0)

Using a compiler flag (such as GCC's __attribute__((packed))) to fully pack a structure removes all alignment padding and so reduces the size in memory and cache usage.

However, as mentioned, on some architectures attempting to read/write things bigger than a byte non-aligned causes a fault which can abort the program (or hard fault the CPU in the kernel or when using embedded bare metal or RTOS usually causing a reboot).

Some CPUs allow an excepton handler to analyse the fault and pull/put the bytes out/in one by one so that the user's assembler or higher level code doesn't have to worry about it but that's going to be very CPU intensive itself and may reset the instruction pipelining and caching.

Some architectures handle unaligned access transparently entirely within the hardware but there is genernally a bus cycle penalty if two 32 or 64 words have to be read or written across the memory bus and again this can cause stalls in the hardware optimizers.

So it's a good idea to have an understanding of your hardware platform even when writing apps in userland on a POSIX, Windows or other high level OS.

Atlassian's Bitbucket Cloud went down 'hard' today

Re: > The om;y worse set of tools....

You've not used Gerrit, then?

Oh my. I'd forgotten about (or possibly blanked the memories of) Gerrit.

K.I.S.S.

I am repeatedly baffled as to why anyone chooses to pay for Atlassian product. At first glance the web UIs look very shiny but once you start to use them you realize how disfunctional they are. Bitbucket provides nothing that services like Github or others don't. Confluence is a piss poor wiki/colab/CMS compared to many others. And Jira, well Jira, is possibly the most inflexible, sluggish, non-intuitive issue tracker I've used. Even Mantis & Bugzilla are better and Jira doesn't even come close to addressing the sort of peer review that should be de rigueur these days let alone safety or security critical development regulatory requirements.

There are open source integrated systems that provide the same functionality available online, self-hosted or hybrid. In the past I've rolled more effective systems piecemeal from the likes of Subversion, Mantis et al.

The only worse sets of tools I've used are PTC's (Windchill PLM ...) and LDRA's code analysis, especially given the eye-watering prices they charge.

Million GPU clusters, gigawatts of power – the scale of AI defies logic

What the hell are they going to use to train all these new monster LLMs?

There's already a significant percentage of 'knowledge' on the Internet that has been churned out by the squawking birds.

Even if the current methods could lead to AGI, there's not enough good data upon which to draw, even if using copyrighted material that should be firmly off limits to these parasites. If they're allowed to squirt it back out into the source pool it'll all become worthless to them and the rest of us well before they've got these gas-guzzling things fully online.

It's an informational Ponzi scheme in a similar way to almost all cryptocurrencies being financial ones.

Now I don't even have to mess up Wikipedia articles myself. I can get my LLM to do it.

Open source router firmware project OpenWrt ships its own entirely repairable hardware

Re: Repairable?

As long as nothing is BGA, you can buy a perfectly good SMT rework station from Amazon for less than £90 which includes a soldering iron as well as the temp-controlled SMT hot air gun and a set of tweezers.

Learning how to use it without incinerating the PCB is a little trickier. Practicing on scrap is a good idea and I'm still learning to do devices larger than a 16 lead SOIC!

Australia passes law to keep under-16s off social media – good luck with that, mate

Re: Australia's government counters ... many studies have shown social media is harmful to children

Cite some peer reviewed scientific studies that demostrate actual harm to under eighteens.

All legitimate studies and meta-studies to date find no link. For example, try PubMed.

Twenty years ago, people were writing similar scare stories suggesting a link between video games, mental health and violent crime. Again, upon rigourous analysis no link was found.

I do not like social media and you may not like it either but those are our opinions and we don't get to assert that people of any age shouldn't be using it without actual evidence.