Mig gas bottle.

[birchy]

I guess experience does help, but I don't find differences in using CO2 or Argoshield, other than CO2 is cheaper and gives better penetration. The main downside of CO2 vs Argon mixes is that CO2 can freeze the regulator under heavy use, especially if gas flow is set too high. I suppose gas quality also varies from supplier to supplier and I dunno what those disposables are like....pretty poor I would imagine.
EDIT: I just remembered a conversation I had a few years ago about CO2 quality. The basic conclusion was that "pub gas" CO2 has better purity than fire extinguisher or welding gas because it is "food grade". Whether there's any truth in that, I really don't know. I've only ever used "pub" CO2 bottles on my home welder.

I did take a look at the link posted above, but didn't read beyond the first few lines:

All welds were made in the same hour with the same welder settings

You cannot simply turn the machine on and change the gas without changing settings! No matter how experienced someone is at welding, they'll never produce a decent weld if the machine isn't set up correctly. Every single job needs to be set up to suit the materials being used. Even the gas flow may need to be adjusted as it DOES affect the weld finish. Too much gas flow can be as bad as too little. Another big pitfall of home users is that they buy cheap equipment like those Chinese things or the low powered sets like the Clarke 100A which aren't much good for anything other than tacking sheet metal. Personally, I'd never buy a welder with less than 160A, with 200A being preferred if you can afford it. It's basically down to duty cycles and having plenty of head room so you're not running flat-out all the time. I bought an Oxford 180A mig nearly 20 years ago and have used approx 8 or 9 15Kg spools of 0.8mm wire in that time. The only maintenance I've had to do is blow the dust out a couple of times and replace the gas shroud and a few tips. On certain things, it doesn't pay to buy cheap.

[Musht]

Must admit been subject of some debate and experimentation over the years..

> CO2 can freeze the regulator under heavy use

Few turns of microbore copper wound round a 60W inspection lamp bulb used to be the answer to expensive heated regulators appparently.

>"pub" CO2 bottles

Some `smoothflow` or something , gases have a nitrogen/CO2 mix for smaller bubbles, worth avoiding.

Was one theory about CO2 being `rougher` than argon mixes, was possible lower temperature freezing out more atmospheric moisture into the weld.

Big machines certainly help, being handed torch of an already set up 350A Cebora was a whole lot easier than trying to set up a Migmate 100.

Little bottles main problem is once opened they rarely reseal properley so can cost wrong side of a tenner to make a couple of tacks.

Fabrication welding on the bench with new stock a bit differnt from repair welding of rust to rust in a freezing lane as well ;-)

Wire feed is one thig that sticks in memory as being much steadier on big machines, colleague used to repair machines, Oxford used to make oil-cooled copper cored welders, another world from air cooled aluminium, but another universe price wise as well.