So the way I see it, there are three types of dive computers out there.
A) Those that are designed strictly for the recreational diver: Manufacturer has already determined and algorithm and you can not change that either by purpose or by error. For Suunto and Cressi, it will be RGBM (Reduced Gradient Bubble Model) and Hollis DG series has Buhlmann with predetermined gradient factors. These are the simplest, most user friendly computers because they enable the user to "strap in and jump." There is no chance of user error as the only choice you will ever make on them is putting the nitrox percentage. Examples are, Mares Puck, Cressi Leonardo, Hollis DG 02 etc. For a recreational diver, any computer above these may not offer any advantage in the dive profile itself. You may have to justify the value of higher-end computers on totally different criteria such as larger display screen and better cosmetics.
B) Rec-Tech hybrids: These are designed for technical divers who dive mostly recreational dives but will also do occasional decompression dives. Most of the time, these run the Buhlmann algorithm and you can not change that. They do give you the choice of changing gradient factors while staying within the algorithm and you can also switch gases during the dive for decompression. They bombard the user with a much wider range of choices than the the first one. If you are strictly a recreational diver then all these additional choices that you will see concealed within their roll down menu will not be necessary and may also open chances of human error. A friend of mine borrowed a dive computer from his technical diving buddy and this one had some technical diving presets from the original users technical dives. These were detected and fixed but had he continued diving with them, then the dives would be very conservative. Some purely-recreational computers start beeping alarms the moment you cross into NDL. You can not miss those as they will very loudly let you know that they are angry. These intermediate-range computers make different assumptions about the user. The moment you go into decompression, these computers tend to assume that you are deliberately doing a decompression dive on back gas. Instead of yelling warnings, they will very quietly start displaying how much time you need to decompress on back-gas before you exit the water.
I have a friend of mine who exceeded the decompression limit on a computer during her dive. I am not sure which model was it or if it was even a Tech-rec one. I am quoting the story as it has some relevance. She looked at the decompression time she had accumulated and thought that this is how much time she had LEFT at that depth! It took some time for her to understand that instead of the number coming down this number kept rising and that is when she realized that she was already well into decompression. She got seriously bent after the dive and was treated for a hit. Not all of that could be attributed to this reading error though. There were other errors made during that particular dive but when you are stressed out and under narcosis then little things can very easily compound the situation.
Bottom line is that by throwing more options at the user they also create a greater margin for human error. At times, this could be on the conservative side while on other occasions it could be on the risk side of the spectrum. Examples of such computers are Hollis TX-1 and Dive Rite Niteq etc.
C) Dedicated Technical Computers: These are designed primarily for technical divers. They allow the users to change, not just gradient factors but algorithms themselves. Decompression theory is changing and these dive computers allow a highly informed user to accommodate whatever research is being put out there on a constant basis. Not all in the technical diving and decompression community agree on what constitutes a safer dive. Bubble models like those on Suunto start to diverge from Dissolved Gas models like those on Hollis and Dive Rite. One family of computers will start generating stops much deeper as you come up while the other family of computers will get you to the highest, shallowest part of the dive that it can get you to without bending you. After that it will keep you there for a long time. Which of the two is better? These computers allow technical divers to determine who they trust more, how much risk they are taking under which scientific assumption and how they plan on distributing their risk between these two models.
Examples are: Perdix and Petrel etc.
When these mid-to high-end computers end up in the hands of recreational divers, then they are using them to do exactly what a very cheap and user-friendly Mares Puck does. The only difference is that now, it is not as user friendly anymore. When I am changing from Air to Nitrox on my Mares Puck then it is a one button process and you cant go wrong on it. When I am doing exactly that on my Hollis TX-1 then it is a three button process to navigate through the menu, set the gas percentage and then make it your primary gas out of the 6 gas options. Yes I have dived on the wrong gas, I will admit it openly. It was during practice dives so it was a negligible error but it would not have happened on the low end Puck that I also own. The TX-1 will take a few more steps to eventually become a "Mares Puck." I have both the Puck and the TX-1 and if I was diving strictly recreational limits while staying within NDL, you wont see the TX-1 on me.
As human beings we are all wired differently so a lot of it will come down to personal preferences and there is not single correct answer. A lot of times these high end computers are acquired simply because of better display and not much else. If I was starting out in diving then I would purchase an entry level Cressi Leonardo on wrist. While it does not have an LED mini-monitor, it still has a decent and well lit display. The conditions in which this display is unreadable may be the conditions extremely unsuited for a recreational dive. (Zero vis cave? I am not sure.) It doesn't overwhelm the entry level user with a long menu and the only human error it lends itself to is wrong nitrox percentage. As long as you can make sure that your nitrox levels are correct, you can focus on other things.
As for integrated air, I am not an expert on that subject. The technology has gotten more reliable and we are seeing less failures today than a few years ago. Batteries still keep running out in the middle of worst moments. My own training has been either through DIR instructors or DIR minded instructors and air integration was seen as a "failure point." I see my set up through Murphy's Law and ask myself, what are the failure points in this? If anything that can go wrong will go wrong on this dive trip then what is it? Plastic buckles will break so I have steel d-rings, straps will snap from the connecting point so I have a once-piece harness. Mechanical objects are more reliable than electronic ones, specially in salt water so SPG is more reliable than air integrated but exactly how much more reliable? That is still open to debate and you see that debate happening on scubaboard on other discussion threads.
If I was starting out in diving, I would buy a strap and dive Cresse Leonardo, clip a mechanical SPG to read air pressure and jump in. Let others debate whether mechanical SPGs are more reliable than electronic air-integrated ones because by the time they have settled that debate, we will all be diving rebreathers anyway.