In my campaign I have 8 players. I'm thinking of lowering the point buy from 27 to 15 points (I assume it is the equal of rolling 3d6 like in older editions) to both have a session about an average guys going out on adventure and to avoid the necessity of balancing every encounter with other means.
Would it still be playable with this house rule?
The XP guidance in the DMG already cover how party size factors into things.
If your goal is...
to avoid the necessity of balancing every encounter with other means
...you're doing exactly the opposite.
The encounter system is designed with the standard array in mind. Deliberately weakening characters across the board is going to require you to rebalance everything by eye, rather than using the guidance in the book.
You need to discuss with your group about what kind of game they want to play. I once played a game where all PC were kobolds, everything was way scarier, we died like kobolds and re-rolled characters all the time. If everyone agrees, by all means do it. But do it because that is the game everyone wants to play not because I am not willing to tailor the adventure to the group size.
Game is only fun if everyone is having fun.
All of the challenges in published works assume the PC were created using the standard guidelines. Tweaking them up or down has unforeseen problems down the road, even making some classes more savory than others.
Example:
Comparing a fighter with STR 10 and one with STR 18 versus an AC 16 enemy:
The STR ability score increased the fighter's damage by a factor of 2.83x - this has direct impact on the combat duration and the damage received from the monsters.
The spellcaster comparison will be left as an exercise for the reader.
It won't make a huge difference, the game is still playable, but it will be noticeable to experienced players.
Here's a typical array (before adding bonuses due to character race) from 27 point-buy:
15, 15, 13, 10, 10, 8
Here's an array you could build with 15 points:
13, 13, 10, 10, 9, 8
As a practical matter, such a character will be 1 down on bonuses to prime stat + secondary stat, have no bonus in third-choice stat (where they might of had +1 or +2 - typically adding to hit points or AC), and maybe have two bad stats instead of 1, affecting saving throws. As there will always be some racial bonus, you are looking at the difference on 1st-level character, let's pick a simple Str/Con/Dex fighter using sword/shield and wearing Scale Mail:
Standard point buy = AC 18, HP 13, Attack +5, 1d8+3 damage
Weaker point buy = AC 16, HP 12, Attack +4, 1d8+2 damage
Probably the biggest effect you will see is reduction in hit points and armour class for those classes where there is not immediate benefit to those from their prime stat choices.
5E is not as balance-critical as previous versions, so although these numbers are lower than recommended, I don't see that they break anything. You take a larger risk of accidental character death (lower AC, lower hit points, less chance to hit all equal more chance of being snuffed out by average goblin at low level), but you could see that as part of the feel you are going for - ordinary people forced into dangerous situations are less likely to survive than tougher heroes.
However, I don't see that this has achieved much. Possibly in terms of feel for anyone already used to the system, they will notice that their characters are worse than usual, and that could make them feel more mundane and fragile. In terms of you balancing the game, there is nothing really to base this on, so you would have to play test it and arrive at something that works. It is definitely not as clean and simple as considering the total number of character points used, for example. And it won't do anything about your main problem - practicalities of running a large game with 8 players around the table.
If you do still decide to start this way, purely for the "feel factor", consider granting additional stat boosts - maybe at 2nd and 6th level - so that as the characters become more heroic, they are not held back by the low stats, and the game will move closer to design expectations.
D&D isn't really a good system to RP 'average people go adventuring'. D&D isn't really good at modeling 'average people'. This isn't a 5e thing, this is true for basically every edition of D&D. For some editions, this is a conscious choice; average people are average and thus not really important to the game system. It's assumed things involving average people will be more out-of-dungeon hand-waving than in-dungeon gaming. When they do come up, average people are generally presented as hostages, escorts, people who shouldn't get into trouble but might, expendable torchbearers, that sort of thing. Generally, the only relevant questions are "How many hit points do they have?" (although in very early D&D the answer is 'one', cause that's what a hit-die meant-- how many average-person-deaths a hero could waive) and "What's their AC?". As the system has evolved, 'average' people have come more and more into focus, but the system hasn't really adapted to handle this, except by making all of the characters, on average, exceptional.
It's hard to be 'pretty ok' at something in D&D. Take 'washing dishes' for example. If you're made to roll a check to wash dishes under pressure, or perform any similar routine activity, your chance of failure is coded to a single d20 roll (or 2 if you're lucky). That means you can't be less than 5% likely to succeed without being garunteed to fail, and you can't be more than 95% likely to succeed without being 100% likely to succeed (or, with advantage/disadvantage .25%, which is better). 'Average activities' are a case where this gradation breaks down-- players will notice that 5% (or higher) chance, even if no failures actually occur, and it impacts the experience of the character. This can be mitigated by ruling 'failure' on such checks to constitute mundane failure, like taking a little longer, rather than exceptional failure, like breaking the dishes or being unable to complete the task, but that isn't how check failure normally plays out, so it comes off as kind of weird.
Furthermore, characters' actual failure rates for mundane tasks at low levels is already weirdly high. Consider a character making use of a non-weapon proficiency which the have and for which they possess a +3 modifier from the corresponding attribute. Some fairly reasonable, mundane tasks might be given a DC of 10. When trying to perform such a task under pressure, said character (who is much better at this task than most 1st level characters and FAR better than your suggested PCs) has about a 15% failure chance without advantage. Even with advantage, that character has a 3% failure chance. If I failed to notice red stoplights 3% of the time I was a little distracted, I would be in some serious legal trouble. I wouldn't be thought of as an average driver, let alone one of the best human drivers possible. And yet that same PC has a 10% chance of succeeding at DC 25 checks. And a character with merely one point less in the relevant attribute score has half the chance of such extreme checks.
In any case, my point is this: by lowering ability scores across the board, you don't make PCs that are okay at stuff instead of exceptionally good; you make PCs that are exceptionally bad at stuff instead of exceptionally good. That can be fun (as another answer mentions), but it's not what you say you are wanting to do.
What I'd recommend is that you try out a different system that is intended to work well for the 'average people take up adventuring' thing.
One of the biggest achievements of 5e is reaching a balance between classes, unlike 3e and 2e, where spellcasters started much weaker, and ended up much stronger than weapon users.
You would end this balance, in an even worse way. As weapon users rely more on ability scores, they would become weaker on every level.
For a fighter 14 Str instead of 16 is a 15-20% decrease in DPR.
For a Fireball 14 Int instead of 16 is not even 3%, as it still deals damage on a save.
Monks need two good ability scores for decent AC alone, and lacking good ranged options, can't dump Con.
tl;dr: don't do it
Just start playing, the problem solves itself. With eight players every encounter (not just combat) will be excruciatingly slow, in a few sessions enough people will stay away.
Or just split the group in half.
TL;DR: It won't change combat much. It will make non-combat less interesting.
Obviously, it depends on the group. Are all your players trying to get the best numeric advantage, etc?
Let's assume they are (because that makes math easier) A standard 27 point buy is likely to lead to a 15, 15, 15, 8, 8, 8, or 15, 15, 13, 10, 10, 8 array (or similar). Typically DnD 5e characters have a stat they really want to boost (Str. for a non-finessing melee type, Casting stat for spellcasters, etc.) That's not going to change.
However, you have a lot less freedom in a 15 point buy. I imagine every character (again, trying to maximize effectiveness in encounters) will be 14, 12, 12, 8, 8, 8. Because the +1's in your secondary attributes (e.g. con) are worth more than avoiding a negative in a tertiary attribute, and after the 14 its all 1 point for +1 stat. And there's no point to odd stats unless you have a +1 from race. If they are playing a race where they get a +1 to the main stat, it would be 15, 12, 10, 8, 8, 8.
So, your characters will do as much damage as reliably with their main attack. They will have either a lower AC or HP/level by 1 (assuming the secondary stat is Dex or Con, which it usually is). Half of their saves will be one lower (their 2nd, 4th and 5th favorite stat). And the characters will feel a lot more similar, because where you're losing the customization is mostly felt in which stats they want to be 4th and 5th.
It also will either (a) prevent anyone from multiclassing (only one stat above 13!) or (b) make you less effective (by 1 to your modifier) while multiclassing (using a 13, 13, 13, 8, 8, 8 array) unless doing something like bard/sorcerer where the same stat is the "main" driver of the class. This 1 point will probably make the multiclass character less effective over the course of the game, but probably won't be noticed at any point in time. However, this player may notice that he misses a lot of attack roles by 1/a lot of enemies make saves by 1 and feel a upset because other players didn't get that decrease to their main stat when you decreased the stat buy.
Also, if you are playing with feats, you have to think about this discouraging people from buying the +1 stat feats. Those are great if you have an odd stat already because you get the numeric +1 to the mod a cool ability. But if you push your players to 14, 12, 12, 8, 8, 8 those feats cost a lot more.
