BS: Micro$oft shapes genetics !!!!

Subject: BS: Micro$oft shapes genetics !!!!
From: Mr Red
Date: 11 Aug 20 - 04:41 AM

Not exactly a gotcha.

Sometimes it’s easier to rewrite genetics than update Excel

Basically the story is that Excel can read genes as a date. So it is easier to change the (internationally recognised) name of a gene than to stop Excel automatically seeing anything (free form) as a date and insisting on formatting/changing it thus.

A small but visible reminder how AI can get it wrong, except this was a human decision, long established.

Subject: RE: BS: Micro$oft shapes genetics !!!!
From: Donuel
Date: 11 Aug 20 - 06:17 AM

Whats in a name? Apparently a helluva lot to AI

Subject: RE: BS: Micro$oft shapes genetics !!!!
From: Charmion
Date: 11 Aug 20 - 08:19 AM

Is Excel compulsory? Surely it's not the only spreadsheet application on the planet?!

Subject: RE: BS: Micro$oft shapes genetics !!!!
From: DMcG
Date: 11 Aug 20 - 09:00 AM

I have had a lot of similar problems over the years. For example, phone records may include IMEI and IMSI numbers, which are 14 or 15 digits. Unless you take special care, Excel will turn these into, say, 1.234E+14, certainly losing precision on the display, and potentially in the number itself. While you can take care with your data set, you cannot be sure if everyone else is taking the same care. Someone who uses your outputs, for example. I had a similar effect with invoices numbers on another occasion. Things that happened to resemble dates got converted to dates, all the other data in the column remained as strings.

Excel is definitely not the only spreadsheet tool, but all have similar quirks in my experience. Converting what is typically pure text data (such as a CSV file) into 'typed' data is fraught with difficulties. As for people putting things like addresses into CSV format: hours for fun when one of the fields has a comma in the value. I have fought many a campaign to get people to use tab separated data instead of comma separated, with limited success.

Subject: RE: BS: Micro$oft shapes genetics !!!!
From: DMcG
Date: 11 Aug 20 - 09:22 AM

I should also have said it is not just what spreadsheet tool you choose to use: if you pass data on, it is also about what tools every one of your consumers use.

Subject: RE: BS: Micro$oft shapes genetics !!!!
From: Mrrzy
Date: 11 Aug 20 - 09:29 AM

So remember Hugh Grant and the woman named Divine? Apparently Grant set Bill Gates up on a date with her, and "after" Gates said Wow, I can see where you got the name Divine. She answered yeah, and I can see where *you* got the name Microsoft!

Subject: RE: BS: Micro$oft shapes genetics !!!!
From: DaveRo
Date: 11 Aug 20 - 01:08 PM

Excel copied its date processing from Lotus 1-2-3, which was the dominant spreadsheet in the '90s. Excel also copied - deliberately - a bug in Lotus 1-2-3 which assumed 1900 was a leap year.

Lotus 1-2-3 copied a lot of features from SuperCalc (I don't remember whether SuperCalc automatically formatted dates), and SuperCalc copied a lot of features from VisiCalc, which was the first spreadsheet.

A lot of features which you might associate with MicroSoft Office go back to the '80s or earlier.

Subject: RE: BS: Micro$oft shapes genetics !!!!
From: Mr Red
Date: 11 Aug 20 - 01:35 PM

Is Excel compulsory? Surely it's not the only spreadsheet application on the planet?!

No, but it is ubiquitous. It is not just scientists that use it. Journalists, statisticians, governments, politicians (who are anything you want, but not tech savvy!).

And who would think when importing 1000 lines from, say, a CSV, to check each line and column? Especially if you didn't know the problem existed.

And with increasing dependence on personalised (gene) therapies - this is life threatening.

Remember the Y2K problem? People blamed the computers, but it was mostly a legacy for when people understood the shorthand for dates. Which predated computers. Like 102 year-olds getting medication for 2 year-olds. When the 102 year-olds were 2 years old there were precious few 102 year-olds. Needs change!

Subject: RE: BS: Micro$oft shapes genetics !!!!
From: leeneia
Date: 12 Aug 20 - 12:35 PM

Microsoft isn't shaping genetics; Excel is affecting how genes are named. Big deal. Actually, not giving a gene a name that looks like a date sounds like a good idea.

Subject: RE: BS: Micro$oft shapes genetics !!!!
From: robomatic
Date: 13 Aug 20 - 03:21 AM

This is reminiscent of the apocryphal tale of modern rail gauge descending from Roman chariot wheelbase.

I love it.

And there is of course the even more wonderful tale of the mucus in our noses being genetic descendants from worm goo a billion years ago.

Subject: RE: BS: Micro$oft shapes genetics !!!!
From: Mr Red
Date: 13 Aug 20 - 03:54 AM

FWIW - (when I remember to be annoyed) I set all fields to text when importing data from CSV. And the same goes for splitting data.

But should all columns be in text mode beforehand? - I am not sure.

BTW there is a second auto-format Excel effect with engineering notation. Genes given , say, 23E-1 (or some such) can be translated to 0.23.