Re: Lower ranking officers
Private Browsing says... Hold my beer - SAH!
116 publicly visible posts • joined 12 May 2010
To be fair, the NAO arent chuffed with the situation and having had interactions with them do have suggested pragmatic governance in publically available documents, I thought good observations on the problems and how to address them. But no-one wants to listen to 'can we just have a think about this first'. See NHS IT scandal and the yet to occur AI everywhere nonsense announced recently.
We work in the 'is this technology worth it' space and determine cost / benefits, impacts and issues - in great human fashion, no 'leader' wants to hear that their idea or concept has downsides or won't deliver all that was promised. Progress is measured in how much physical stuff has been visibly bought, delivered / installed, not in whether there's been upfront work to understand and de-risk the project. By the time it starts going wrong, the original 'leader' has moved on, usually promoted.
The Blair Institute has been peddling this for the NHS and now AI everywhere! What the AI purveyors wont say and the desperate (in this case the Government) cant acknowledge is that pretty much 'AI' is used when you need to make a decision (do I do this or that). Before implementing AI you need to understand what decision you're trying to improve, how much you can improve it, whats the cost of using AI to do this AND remain with legislative frameworks (from use of training data to explainability as in the GDPR). You also need a stack-ton of supporting infrastructure and clear understanding of your processes.
Wishing on a star is going to be an expensive exercise in buying toys that get left on the shelf, but yes, buy that Palantir stock now!
Data has no value. Information extracted from data has value. Just storing data and hoping for the best is bottom-up nonsense. You should understand what problem you're trying to solve, from this derive an approach and the data (and its quality) you need to support that. Data supports something you specifically want to do, its not an end in itself.
Part of my life is involved in finding issues with the incorporation of new technology to upgrade or 'improve' things, prior to bringing into service. There are always a bunch of knock-on issues related to through-life operations to the shiny new, some of them so significant that they result in major changes to the solution or in several cases the solution is dropped. I see both sides here, yes Rust 'can' improve certain things and if this was a green-field project, fantastic. Unfortunately there are existing personnel, skill-sets, processes, legacy code-bases etc that the shiny new has to fit into nicely.
It's very common for tech providers to go 'I've solved all of your problems!' and for the stakeholders to go... 'You do know how things work around here, right???'
Agreed on the suicide mission bit, Shunsuke Baba is head of Service Technology so it looks like Fujitsu has Fujitsu as a customer. First thing I thought when I saw this was... do they never learn?
Sorting out business processes always looks easy on visio but is a pain in reality - as ye' ol' saying goes.. In theory, there is no difference between theory and practice - but in practice there is.'
Machine learning needs realistic training data to perform a classification task, which this is - is the data presented a 'technosignature' or not. Ideally you'd like a classifier that says 'I've seen something anomalous or different to what I've seen' but that's not how classifiers work, you do need representative data for each of the classes. (Humans are really good at noticing when things are odd without representative data, ML classifiers are not). There is no representative data for a technosignature so its a complete guess if one makes it artificially. This guess is driven into the classifier.
So interesting work and a great idea for a 3rd year undergrad project where it's really more about the project than the discovery, but this isn't serious analysis of 'its aliens'!
Its been interesting how the hungrier AMD has caught up with Intel - at one point it looked like they were going to the wall and failing. I was told a long time ago that people work best when they're 'hungry', before anyone gets narky about pay and literal hunger, the comment meant not having gazillions in the company bank account that you can free-wheel on and hungry to prove you can do it. Intel have been fantastic over the decades but someone always catches you up or chases an innovation you didn't think to pursue.
The best description I've heard of today's 'AI' is just a jumped up autocomplete. It needs humans to curate all the data, humans to hand fettle, design the structure and build the transform in the network (for that is all it is).
The excitement over GPT3 is just the latest incarnation of people getting excited over implementing tech (and the notion that crude approximations of networked nodes is clever like a brain as it's structured like a brain), as opposed to why the tech solves a given problem. A while ago it was things like Alexa. Before that it was apps on phones, Before that it was Victorian clockwork automaton (as back then the world was deterministic).
The thing about GPT3 and its ilk is that as a thing it is clever, however its output isn't.
You're dismissing poor Chinese build standards as nonsense!? Have you not seen or researched anything regarding the corruption and poor standards in China - its pretty everywhere?!
HSR is not one jot near the complexity of a modern airliner. Certifying an airliner is one thing (and they have struggled badly for this), creating a competitive airliner is another order of magnitude of complexity again. There is every reason to believe that China will never compete with Airbus / Boeing. That demo plane has sat rusting for years as China has struggled to develop it. I've chatted with some European consultants (useful idiots) that China bought in to help develop the plane and its clear that China has neither have the skills or the mindset.
Oh and HSR was not developed by China, reading the same Wikipedia page you did shows that they were built using technology transfer agreements from world-class market leading train makers. You'll also see that its a commercial disaster as well so its not going to wear out soon.
Collecting, curating and labelling speech for part of a speech recognition database is time-consuming, difficult and expensive. Depending on the speech disability one might also need to re-specify the dictionaries as well as the acoustic models may not match. It might also need differing feature generation as (for example) those with speech disabilities can struggle with plosives (so less emphasis on consonants).
I agree totally that we should focus on enabling technology, the issue is its not always straightforward to map the current approaches to optimally support someone with a specific disability. Obtaining data, either a complete data set or using adaptation data is time consuming and hard work.
It's not AI.
It's just a transform from input to output. Potentially complex from a human perspective but just a transform. It models and represents only aspects of the training data.
Personally I wonder about this, the researchers should understand the provenance, validity and scope of their data. If they're just using someone else's scripts to train a model (eg lots of 'researchers' use YOLO for image recognition as it has easy to use training scripts - but don't really know how it works ) then that is poor.
So researcher doesn't know exactly what they're doing, identifies biases and thinks that is indicative of racism / sexism...
No, this is quite poor and indicates a lack of understanding of machine learning and statistics.
OK, yeah, I can see both sides of the argument here, but personally I like the transport layer to be not closely coupled with the application as this is a bad idea long-term. Whenever different system elements are closely coupled to give an integrated improvement in performance, that's good for today and less so for all of the tomorrows. Through-life support and all that.
Plus, as I've got older, the first and second rules of optimisation make more and more sense.
These rules have never worked. HMRC is out of control and has a vendetta against small businesses. I don't know if its because it makes their life easier if you just have a few big businesses you can have cosy chats with or because they've been given advice by companies that would like small businesses out of the way.
HMRC refused to appear in front of the Commons Business Committee when they were investigating IR35 - HMRC are big on kicking the little (man || woman) but don't like answering questions regarding their own governance.
Yeah this is all a bit confusing. Terra Quantum have got 75million which they say will be used to strengthen their crypto{graphy,security} offerings...
But then the article mentions improving chip design by the use of negative capacitance (snigger, inductance!) to counter the built-in (positive) capacitance across the gates in MOSFETs which messes things up. So my intuitive analogue electronics take on this is this then forms a tuned resonant circuit which just creates a whole other host of practical problems. (For one if you're not switching the transistor at its resonant frequency the (positive) capacitance or (negative (snigger, inductance!)) dominate and you're back where you started. Plus in my experience resonant circuits tend to ring like little sods when you'd prefer they shut-up - especially when getting slapped with a nice impulse.
Thankfully the papers rapidly descend into the use of quantum speak making it impossible to argue against...
It's interesting to hear what the younger 'woke' generation think is 'nice'. There's an organisation you need to hear about called Amnesty International. They have some quite good examples of what 'not nice' is.
https://www.amnesty.org/en/latest/ (It's in large fonts for the hard of thinking too.)
You might realise what a lovely fluffy bunny world you live in, compared to, well most other places. And how you would be treated if you visited there...
No, because you pay Corporation Tax on dividends (as well as income tax) which fluctuates around the 20% mark (currently 19%) so in the grand scheme of things it is pennies.
If you're a contractor believe me you feel like you're paying tax all end up, mainly because you are.
It is a fantastic piece of free software with some very clever algorithms and plug-ins. A lot of comments relate to it being difficult and a steep learning curve and that's fair, but complex things are hard to learn - personally I've plugged away and got it to do what I want, not always on first go.
Complex things either prescribe how you use them, in which case its a pain to get them to do anything different from the author's intended approach, or they provide a tool-kit of options that you have to figure out, gives that steep curve, but is far more flexible. For beginners, gimp and photoshop are similarly complex...
"Not an American? Then mind your own f*cking business."
Correct - this is not an American web-site - it's quite worldly. You have a lot of anger, some of which is understandable, but rant somewhere relevant or cool off as you're doing no-one any favours.
On a more relevant note, if anyone is interested in magnetars this is quite an entertaining 15 min explanation (I have no connection to this, I just enjoy his videos).
https://www.youtube.com/watch?v=K8sCG0rJitI
The tragedy of the commons....
190000kWh is a staggering amount of energy that doesn't produce anything tangible. If people using gpt-3 or any other ai are 'clever' then they should be responsible enough to account for the impact of what they're doing.
I've read an estimate that data centres account for 3% of global emissions - the same as air travel. Yep there are error bars there but these things have a cost that needs some recognition.
1. The supposed increase in performance uses a different data set. You cannot claim an increase in performance for your technique if you use different data.
2. You also cannot claim an increase in performance if the false positives aren't stated - just saying everyone is of class X, correctly identifies everyone may be in class X. They may not have Alzheimers or they may have something else as Screwed rightly points out.
"Get a job with a Company" - This isn't Russia - why should my small Company be picked on to the benefit of the Qinetiqs, PwC, KPMG, Leidos et al?
"Put your rates up" - thanks, hadn't thought of that. The problem with the Putin tax is it is so swingeing you cannot commercially put rates up high enough to compensate.
"Work for more than 2 people per month" - you haven't a clue about IR35, have you? Its on a role by role basis. You can work for one day and still be within IR35.
"The issue is, you're effectively an employee" - if I'm an employee, then EVERYONE who contracts to another company is an employee, so EVERYONE should be caught. Defence Prime supplying the MoD? Well, you're acting as pseudo-employees and working under supervision - day-rates get hit by 63% tax. You provide a service contract? Employee by definition - bang 63% on your service charges. You do design? Then you're acting as a pseudo employee design team - 63% off of your company charged day-rate. Then take it out of the employee wages - maybe that will focus a few minds which are coming over as a bit blunt.
In the last 10 years, there have been 16 IR35 court cases that HMRC have instigated. Just 2 were won by HMRC, and 2 were split decisions. In 75% of the cases that HMRC took to Court, they were proved wrong. Any other body that was wrong 75% of the time would be deemed as unfit for purpose.
This doesn't include where they've investigated, shown a contractor is outside of IR35 and not pursued legal action - I know of a fair few of these.
Deliberately and spitefully manipulating the system so Contractees force a punitive tax regime on Contractors shows the HMRC is rotten to the core. Virtually all contracts have become IR35, not because they are, but because of a fear of HMRC investigation and that's not something fit for a democracy.
https://www.contractorcalculator.co.uk/ir35_court_cases_judgments.aspx
Er - you might want to check your maths - in the last 10 years, when there's been better legal understanding of IR35 legislation the HMRC have chased up just SEVENTEEN businesses, won 2 judgements and had 2 split decisions. Splitting the draws this gives about 84% in favour of contracts NOT BEING IN IR35.
Remember HMRC will only go to court if they think they've got a case, I know a fair few people who have been investigated for IR35 compliance and found outside - these figures do not include those and are not publicised by HMRC either.
"The clients have no one with my skills in the company, so IR35 does not apply."
You plainly do not have a clue. This has never been a determination for IR35. Substitution and MOO have been the main discriminators.
Unless this nonsense is sorted industry wide you'll be hounded out soon. Maybe you'll last longer or maybe you won't, but the doors have slammed shut and no-one is listening.
If you're in a position today where you can decide on your clients, then woop-de-doo, but make the most of it because this industry changes fast and so will your prospects.
That's a very good point and soon Contractors within IR35 will be suing Agencies and Contractees left, right and centre - the first time someone is seriously ill and needs a substantial time off (or paternity leave, or even bank holidays) there'll be legal action, someone other than contractors are going to get stung and this thing will either be dropped or a dose of realism put in place.
Well I'm waiting for the determination from BAe at this very moment. The agent won't push as he's scared of upsetting them. HMRC have made sure that there is no incentive for Contractees to come up with an accurate determination, if you put everyone inside you're fine, there is no punishment if you falsely over-tax - and what exactly is the path for Contractors to get a review? There isn't one. This is not how a democracy works. This is not how a fit-for-purpose HMRC would work.
Still, if HMRC go after the PwCs, Andersons, Mckinseys, Sercos next (a service contract must surely be disguised employment), plus they tax Amazon, Facebook and Apple at 65% of turnover then at least all things will be equal.
I've heard so many employees think IR35 is good (taking contractors down a peg) as they have overpaid rates. The simple fact is its a dumb arbitrary rule to extort cash from small businesses. You go to somewhere like NATS where contractors work identically to say Leidos staff (but contracted in to NATS and answering to NATS staff) workers. Do the revenue bust Leidos' day-rate? Of course they don't. Why not? Explain that little conundrum.
The day the Revenue chase PwC, Fujitsu, Leidos, BT or any other big subby that puts someone on a desk doing what is an employee role, then we can chat.
Contracting is a completely different mind-set and set of skills to being an employee. Remember, contractors are there because your company cannot cope without them. The moment they can, we're off.
About neural networks, what they are, what / how they 'learn', what they do and what the limits of their performance are. I really thought this was all understood 25 years ago, but a new generation has re-sprung nnets into magic and 're-discovering' basic properties.
And while we're at it 'researchers' should also understand how the front-end features contribute to the overall classifier performance too.
Someone mentioned on el Reg about a week ago a cookie delete add-on for Firefox which I then started using. It deletes cookies 15 seconds after you close the tab. That sorts out cross-site advertising.
What is amazing is some sites can have upto 250 cookies tracking your activities and the moment GDPR kicks in I shall be having a word as to why a site needs to inform 250 third parties about your activities...
and you're left to figure out the questions, ignore them.
Trying to come up with useful emergent properties of things from random observations is a fools errand.
This applies to monitoring (including surveillance data), EA models, SysML models, endless cyber-security SIEM 'notifications' etc etc etc...
Anyone that threatens to diminish European harmonisation by a 'European' institution will have their arm twisted by the EU. So yes, they will have something to do with it.
Unification only goes one way and the EU will not have anything that looks like a fracturing of European unity.
You really think that you have an option on this?
Vote the wrong way and your friendly EU will twist your arm until you come up with the 'right' answer. It'll impose punitive measures on your resistance until you toe the line. Ireland has done quite nicely out of the EU so I cant see your nation rocking the boat anytime soon.