View previous topic | View next topic

How has "algorithm" become a dirty word?

Page 1 of 1

Brock
1375697.  Fri Feb 26, 2021 11:10 am Reply with quote

A thought inspired by post 1375622:

suze wrote:
It's a fair point, but do you remember "no deal is better than a bad deal"? No algorithm is similarly better than a bad algorithm.


How has "algorithm" become a dirty word?

In its broadest sense, an algorithm is simply "a step-by-step procedure for solving a problem or accomplishing some end" (Merriam-Webster). A recipe or a knitting-pattern could be described as a type of algorithm, though in practice the term is generally restricted to mathematics and computer science.

It's hardly a new concept. The oldest algorithms, such as division algorithms, are said to date back to 2500 BC. The term "algorithm" itself comes from the 9th-century mathematician Muḥammad ibn Mūsā al-Khwārizmī, although it didn't take on its modern English meaning until the late 19th century.

Algorithms are essential to the way computers process data; indeed, any computer program can be seen as an implementation of an algorithm. Algorithms can be represented as flowcharts, pseudocode or in many other types of notation, but programming languages have the special property of being machine-readable; if you code up an algorithm as a computer program, it can be executed automatically. Algorithms are the bread-and-butter of computer programming, and computers wouldn't be able to function without them.

You wouldn't think any of this from the way the media have seized on the word "algorithm". Up until about ten years ago, the word was scarcely heard outside technical contexts. Then, for some reason, there was a sudden explosion of reports about how computer programmers were now using "algorithms" to decide everything, as if they'd only just been invented. The impression given was that they were something very complex and technical that ordinary people couldn't understand - although it mainly seemed to be journalists and news presenters who had trouble with the concept.

And now they're being blamed for everything that goes wrong. We've just heard a promise from the Education Secretary that no algorithms will be used to assess grades this year, as though it was the very fact of using an algorithm last year that caused problems, rather than any suggestion that the algorithm was badly designed, or that he didn't understand what it was meant to do.

It can't be long before the general public gains the impression that "algorithms" are a universally bad thing, whereas in truth, none of today's technology would exist without them. Will mathematicians and computer scientists have to change their terminology to avoid giving the impression that they're up to no good?

 
barbados
1375700.  Fri Feb 26, 2021 11:25 am Reply with quote

The trouble with the particular process in establishing exam grades is there are losers to the scheme - some teachers wouldn’t have inflated the grade, and the child really would have got the A* rather than the B awarded, and they would rightly shout a lot louder than the child who was lucky to go to a high achieving school where the teacher thought they should get a B and they were awarded the A*
The system was never going to please everyone, and thatis what caused the problem. The day they come up with a universal formula where everyone is treated both subjectively and equally will be the day that hell freezes over.

 
crissdee
1375701.  Fri Feb 26, 2021 11:35 am Reply with quote

In fairness, I think that the fact that so many of the algorithms we encounter in our day-to-day lives on line are near useless has contributed to this attitude. I am registered with a number of job seeking sites as looking for warehouse work within 10 miles of Builth Wells. One has repeatedly sent me details of warehouse work in Essex, and today, when I tried to get it to recognise where I actually lived, it came up with "39 warehouse operative jobs near Builth Wells". Of that 39, approximately* 39 of them were for taking online surveys at home.

As for the Ebay algorithm which works on the principle that having bought a suspension part for my Saab, I must be interested in collecting suspension parts in general, and that searching for Harley Davidson motorcycles means I am interested in anything that has two wheels and an engine, because they're all the same really, I despair of it ever coming up with anything useful.

Examples;











*well, exactly 39 really....

 
Brock
1375702.  Fri Feb 26, 2021 11:51 am Reply with quote

crissdee wrote:
In fairness, I think that the fact that so many of the algorithms we encounter in our day-to-day lives on line are near useless has contributed to this attitude.


Are they? If most online algorithms were useless, you wouldn't be able to log on, because the software wouldn't work. Indeed, you wouldn't have succeeded in posting the above message.

Quote:
I am registered with a number of job seeking sites as looking for warehouse work within 10 miles of Builth Wells. One has repeatedly sent me details of warehouse work in Essex, and today, when I tried to get it to recognise where I actually lived, it came up with "39 warehouse operative jobs near Builth Wells". Of that 39, approximately* 39 of them were for taking online surveys at home.


But that's not a problem with algorithms per se. That's a problem with a particular type of algorithm - artificial intelligence algorithms, the type that allow computers to make decisions that would previously have been made by human beings. It's put quite well in this Wired article:

"The questions being raised about algorithms at the moment are not about algorithms per se, but about the way society is structured with regard to data use and data privacy."

I suppose I've resigned myself to the use of "algorithm" in the media to mean "AI algorithm", but really they're doing computer scientists a disservice. The vast majority of algorithms that are used in computing have nothing to do with AI at all.

 
barbados
1375703.  Fri Feb 26, 2021 11:59 am Reply with quote

That question is one similar to “what’s the point of voting, politics doesnt affect me”
As you say there are an uncountable number of algorithms that affect our daily life - but we don’t notice them because they don’t do anything tangible. The ones that do affect directly have winners and losers, so if you lose as a result of the algorithm, it’s the fault of the algorithm. But all the algorithm does is act on the data in an unemotional way. Then because someone loses out, the media get hold of it and then algorithm = bad

 
suze
1375706.  Fri Feb 26, 2021 12:46 pm Reply with quote

Brock wrote:
We've just heard a promise from the Education Secretary that no algorithms will be used to assess grades this year, as though it was the very fact of using an algorithm last year that caused problems, rather than any suggestion that the algorithm was badly designed, or that he didn't understand what it was meant to do.

It can't be long before the general public gains the impression that "algorithms" are a universally bad thing, whereas in truth, none of today's technology would exist without them. Will mathematicians and computer scientists have to change their terminology to avoid giving the impression that they're up to no good?


They may, and it will be almost entirely Mr Williamson's fault.

There is supposed to be an official inquiry pending into just what went wrong with the particular algorithm, although it may have been quietly canned. That there would be one was announced within days, but no word since.

But it is necessarily one of three things.

1. Ofqual screwed up, and delivered an algorithm that was not as specified. This is what the government wants us to believe, and it is perhaps supported by a couple of the top people at Ofqual resigning. Well, if "resigning" is really the correct word for being asked to leave in return for a seven digit number of public pounds. If I screwed up that massively, I'd be fired without a penny.

2. Ofqual's algorithm was exactly as specified, but Mr Williamson's specification was wrong. But it can't be that, or surely he would he have resigned or been fired. Oh no, hold on, he voted Leave so he has a job for life.

3. Ofqual's algorithm was both as specified and was correctly specified, but the government got scared when the Daily Mail didn't like it. If it's that, I might once again question why we allow a far right newspaper read only by morons to determine government policy.

 
Brock
1376708.  Fri Mar 12, 2021 1:04 pm Reply with quote

Here's someone else who clearly doesn't know what "algorithm" means:

"Bitcoin mining – the process in which a bitcoin is awarded to a computer that solves a complex series of algorithms – is a deeply energy-intensive process."

https://www.theguardian.com/technology/2021/feb/27/bitcoin-mining-electricity-use-environmental-impact?utm_source=pocket-newtab-global-en-GB

You don't "solve" algorithms. Like many other people, I don't pretend to understand how bitcoin works, but that statement is simply nonsense.

 
Awitt
1376766.  Sat Mar 13, 2021 5:44 am Reply with quote

One that comes to my mind is the algorithm the Australian government implemented to check people on income support payments were reporting the correct amounts from any jobs they had during the year, but it is now found to be wrong and people are being paid back.

This program took the amounts people had reported in a year and averaged it over the 26 reporting fortnights, then informed people they have a debt for part of the year.

When in fact they reported correctly in the times they may have had work, and reporting zero income earned when they did not. While I haven't been affected by this, (thank goodness) I reported over $2000 sometimes in the 2019-20 financial year, then zero from last April when Covid 19 hit our shores and schools were closed.

 

Page 1 of 1

All times are GMT - 5 Hours


Display posts from previous:   

Search Search Forums

Powered by phpBB © 2001, 2002 phpBB Group