Anyone with an engineering degree can call themselves an engineer, if not you are a technician.
When I worked for a German company they used to get most upset at us calling all techies “engineers”
The French were the same as well.
I guess us Brits don’t give a shit as I’ve met a few washing machine repair engineers.
I agree about the naming too. But the problem with software goes even deeper than that. The thing itself is the problem. I’m sure Microsoft employ some very, very, smart people. Mrs VB is pretty smart. But they’re all working in an environment that has evolved to the point where even if you do a fantastic job your product will be launched into a sea of crap. Unless you write the ones and zeroes yourself (I once wrote some machine code, which is the next nearest thing) you always have to depend on other people’s stuff. And, even if you can suss that out and build a bombproof product that copes with with all the bugs in it, these days it seems it’s the convention that they’re allowed to creep in when you’re not looking and fuck with their bit (it’s called ‘upgrading’, which would be funny if the consequences weren’t so sad so often).
I don’t know how to fix this situation. In fact I’m afraid the problem might be contagious. Twenty years ago I could buy white goods for the kitchen which would last at least 20 years (I did, and they’re still working). But not now, apparently. The new stuff does at least work out of the box on Day 1 though. I wonder how long it’ll be before manufacturers start excusing failures by blaming the water supply, or the soap powder people …
VB
term engineer has never been a ‘protected’ title in this country. Accreditation is one way of preserving kite marks, but unfortunately almost every sector of industry I can think of are not interested, except perhaps where things are high risk and safety critical or co-terminus with a licence to practice.
I have spent an awful long time working with and for the Engineering Council. These days you can get full CEng accreditation for being a software engineer. You may not personally agree, but properly trained software engineers well deserve the title engineer.
Obviously, she married you 
You can also be accredited as an Incorporated engineer without having a degree, I am. It is also true that the term Technician Engineer is an accredited title with the EC. You would normally be expected to have an HNC/HND. In may countries a Technician Engineer is recognised and respected for the work they do. They fulfill those roles that sit between the spanner monkeys and the Incorporated Engineers, a good one is worth their weight in gold.
there is no fix, the computing industry has always been reliant on the commercial interests of the next person. Even writing the machine code yourself is no good, if you didn’t build all the hardware yourself, and/or write the microcode yourself.
Some very specialist closed systems are still like this, but they are few and far between. There was a thought back in the late 1980s and early 1990s when I was working on the designs for the first quad core microprocessor to go on a single die, along with writing the compiler and crude OS, that the closed system was the way to go to extract every gram of computing power, and to achieve reliability… but the reality is, that still doesn’t work.
Anything that involves errors introduced by humans will suffer in the same way except in computing where it of often (these days) more noticeable. I recall talking to mechanical engineer a few years ago. He had been commissioned to develop a system for precisely cutting a material. He was given a spec, and built a system that worked to spec, 80% of the time - he was satisfied with that and so was his client (as it was within budget and performance spec)… but in computing that same exists, the interdependencies are so great that it is impossible to test and measure for 100% fault coverage within a budget.
The hardware/software (which is multi layered) co-dependencies in a general purpose computing machine are a seriously complex problem. I once wrote a paper on it during my PhD. Oh and my quad core processor - it worked (most of the time), the fab could only get 60% yield, and the built in self test couldn’t test all fault conditions. We only ever found some of the more obscure faults (like too much jitter on one clock line), by using it in the daughter board.
I know
Soz, that was meant to be directed at Chris:)
bloody forum software didn’t take into account the HCI in the design phase
Bloody engineers!
It’s £3, it comes via Microshite and you expect it to be wonderful 
Everything, and I mean everything Microshite touches is crap, really crap. The O/S’s are crap, the systems software is crap, the applications are crap.
Microshite have a policy of beta testing on the paying public/businesses, so what the fuck do you expect?
They employ ‘agents’ to make sure that the first 30 reviews or ‘views’ of their products are 100% positive to get them into the market. Next time you see an article in a national newspaper in the technology section, you will see 30 supportive and great reviews, followed by the actual buyer’s saying it’s shite.
I guess that’s the essence of it.
Your mech eng friend delivered a product which worked imperfectly but acceptably and I imagine he understood why that was and what would need to be done to improve the performance (I’m guessing, but I imagine it wouldn’t have required a complete paradigm shift in his whole subject area). I once built a large-area pulse synchronisation system which had of order 100 femtoseconds timing jitter. Its performance depended on everything - temperature, acoustic noise, the weather, traffic on nearby roads, the phase of the moon (I’m not joking - tidal forces stretched and compressed the building and the ground on which it was sitting which matters given that 100 femtoseconds is the time it takes light to travel 0.03mm). When my colleagues asked if I could make my system work better I could give them a prioritised list of things we could do, with reasonably accurate costs and expected levels of improvement.
In computing, on the other hand, the interdependencies can make the difference between success and complete and utter failure. At first glance Rob’s experience looks like a mixed one. But in a bit more detail we can see that the playing of individual tracks works essentially perfectly (success) but the building of more complex structures - playlists etc - fails almost completely. It’s the regular occurrence of ‘failing almost completely’ which seems pleasingly rare in most fields but depressingly common in software. As for “What can we do to improve this ?” if you can get any answer at all beyond “Turn it off and turn it on again”, it’s likely to involve completely junking either the product itself or the other products with which it’s turned out to be incompatible. Sometimes problems can be fixed by installing an upgrade somewhere. Sadly they seem to be fixed more often by uninstalling one.
VB
I think two of the issues that’s possibly more particular to software than a lot of other technical disciplines is the aforementioned leaning tower of software component dependencies on which almost all software is built, as it’s a case of balancing out time / work required against cost. It’s also the case that whilst there are an array of “standards” for things like computer hardware, there’s a whole mishmash of companies building components that need to work together without them properly talking to each other about producing a coherent system. I think that’s why for such a long time Apple have kept everything on “their” hardware with their software, it hugely reduces the number of permutations you have to deal with.
It’s interesting to see just how much of software design these days is centered around making things “fail gracefully” and with minimum impact to the end user. I think it’s an acknowledgement that it’s probably not possible in the current environment to get close to making things bullet proof, at least not without running up insane costs and in turn having to charge outrageously high prices compared to everyone else. You only have to look at the difference between nVidias consumer and professional cards to see that; for the equivalent raw hardware, you pay many times more for the pro cards, but they’ve put a lot more time into getting the drivers as bug free as possible, and that’s a lot of what you are paying the extra for.
I think the only person who’s ever really consistently put out software that comes close to VB’s engineering standards is Donald Knuth, and he is a certifiable genius by no small measure. Even then, there are still bugs (although they are few and far between). He’s also rather famous for the quote “Beware of bugs in the above code; I have only proved it correct, not tried it”.
See the folks who write the code for fly by wire aircraft, they don’t work for Microsoft.
Much of our dev work has moved to India.
It’s a mess. We tested on scale out and it doesn’t even install. They finally admitted no functional test has been performed at all.
Well, thank fuck that Microshite have bombed in putting their O/S and apps into cars, think of the fucking carnige that would occur 
Especially the spell check
Me.
For mistakenly thinking for all these years that I.T. was dull as fuck, when it turns out it clearly is just the opposite.
indeed as my Mrs (she is an SE, managing about 60 blokes who are Eng, Maths, Phys) pointed out when she read bits this this thread. Certain kinds of SE are only done here by very skilled SE’s. £3 apps for your laptop bear no comparison
yes maybe, there are plenty of others (Wilkes, Dijkstra, Wirth, Hopper are few that come to mind without much thought) mostly all were mathematicians first, computer scientists second… my boss is a personal friend of Knuth…
My dad worked on VME with Bryan Worboys at ICL and then moved over to work on the secure version for defence which he did up until about 4 years ago.
Never earned him any money though as ICL (and then Fujitsu) paid shit salaries, he was happy to stay there for 42 years turning down promotions because he wanted to stay as an engineer.